native Oracle-port on Linux -- what would it take?
S V
Bjorn Borud <bo...@guardian.no> wrote in article
<m2wwh3x5...@lucifer.guardian.no>...
>
> | <sigh> isn't this covered in the Linux FAQ? IF you can get hold of
> | certain SCO binaries you can install Oracle just fine. I have done it
on
> | Red Hat and even on Slackware, both with the iBCS package installed of
> | course.
>
> strange really; it has become a FAQ how to install SCO binaries on
> Linux and quite a few sources say that most large companies or
> companies that have anything to do with the Internet, use Linux -- yet
> Oracle are unable to deliver a Linux version of their database.
Linux is missing many features for any OS to be worth considering
as a viable database platform.
1. Linux has no logical volumes layer.
2. Linux has no transaction-oriented filesystem.
3. Linux has no support for raw devices - hence NO even remote possibility
to run Oracle Parallel Server.
4. Linux networking is flaky at best.
5. Linux OS block size is what? 512 bytes? It would make even MS Access
laugh and puke steam.
6. Linux SMP is rudimentary and flaky at best.
7. The same for multithreading.
8. Oracle requires efficient IPC and record-locking mechanisms. Linux
cannot do
record-locking at all, last time I checked. So Oracle would have to
write its own
code for that. Why would anyone sane want that?
In fact porting Oracle to Linux would require Oracle to write its own Linux
almost
from scratch.
==> I don't see much (or rather _any) prospects for seeing Oracle on
Linux.
Cordially,
Serge
> it probably does make sense when you think about Oracle having a CEO
> that wasn't even aware of what PL/SQL is until someone called him and
> wanted to know if they were continuing to support PL/SQL in future
> releases.
>
> "PL/SQL, what's that!?"
>
> perhaps someone should give said CEO a call and tell him that there's
> this thing called Linux out there that a lot of professionals use and
> that it's time to update one's view of the world a bit?
>
> last time I heard it was the lack of raw-devices, but since that
> isn't really an issue (how many of you use raw-devices anyway? not
> many companies I know) it must be something else.
>
> also software companies keep whining about Linux being a moving
> target. bullshit. pick a distribution and a (release) version and
> make Oracle run on it.
>
> yes, Linux is in flux as are all operating systems. the difference
> mainly being that Linux is evolving and advancing faster and that
> kernel-updates are made available to the public more often.
>
> -Bjørn
> --
> Bjørn Borud <bo...@guardian.no> | "The Net interprets censorship
> <URL:http://www.pvv.unit.no/~borud/> | as damage and routes around it."
> UNIX person, one of "them" | - John Gilmore
>
D. Jeff Dionne
[chop]
> > strange really; it has become a FAQ how to install SCO binaries on
> > Linux and quite a few sources say that most large companies or
> > companies that have anything to do with the Internet, use Linux --
> > yet
> > Oracle are unable to deliver a Linux version of their database.
>
> Linux is missing many features for any OS to be worth considering
> as a viable database platform.
Yet it runs just fine under iBCS emulation on Linux...
> 1. Linux has no logical volumes layer.
> 2. Linux has no transaction-oriented filesystem.
So write support for these, however...
It runs just fine under iBCS emulation on Linux...
> 3. Linux has no support for raw devices - hence NO even remote
> possibility
> to run Oracle Parallel Server.
Have no idea what you mean. raw devices work just fine on Linux,
tar -xvf /dev/fd0 for instance? Works fine. If Oracle needs raw
partitions, no problem.
> 4. Linux networking is flaky at best.
Then the whole internet is flakey, since Linux runs on more servers
on the net than anything else. Rock solid is more like it.
It runs just fine under iBCS emulation on Linux...
> 5. Linux OS block size is what? 512 bytes? It would make even MS
> Access laugh and puke steam.
You're thinking of (an old version of) SunOS :-)
> 6. Linux SMP is rudimentary and flaky at best.
It's actually not that bad in 2.1 I hear, and I fail to see how that
affects the viability of Oracle on Linux.
> 7. The same for multithreading.
But it works fine under iBCS on Linux. Interesting. You're mistaken.
> 8. Oracle requires efficient IPC and record-locking mechanisms. Linux
> cannot do
> record-locking at all, last time I checked. So Oracle would have to
> write its own
> code for that. Why would anyone sane want that?
But it works fine under iBCS on Linux. Interesting. You're mistaken
>
> In fact porting Oracle to Linux would require Oracle to write its own Linux
> almost
> from scratch.
But it works fine under iBCS on Linux. Interesting. You're mistaken
>
> ==> I don't see much (or rather _any) prospects for seeing Oracle on
> Linux.
That may be so, but it has nothing to do with weakness in Linux. I
suspect it has to do with Oracle's relationship with the Big UN*X
Vendors.
>
> Cordially,
> Serge
Arun Sharma
> 1. Linux has no logical volumes layer.
> 2. Linux has no transaction-oriented filesystem.
> 3. Linux has no support for raw devices - hence NO even remote possibility
> to run Oracle Parallel Server.
Don't know enough to comment on those. But you better double check the
validity of your data.
> 4. Linux networking is flaky at best.
And how did you reach that conclusion ?
> 5. Linux OS block size is what? 512 bytes? It would make even MS Access
> laugh and puke steam.
Pure misinformation. The block size is a file system attribute and can
be changed.
> 6. Linux SMP is rudimentary and flaky at best.
And what percentage of database servers run on multiprocessor boxes ?
> 7. The same for multithreading.
Misinformation again. Kernel level multithreading works perfectly fine
with Linuxthreads - a POSIX threads implementation for Linux.
> 8. Oracle requires efficient IPC
Ever checked the lmbench numbers for Linux Vs <your favorite OS here> ?
> and record-locking mechanisms. Linux cannot do
> record-locking at all, last time I checked. So Oracle
> would have to write its own code for that. Why would anyone
> sane want that?
Not reason enough. If this is what is holding up Oracle, Linux
developers can fix it (if it's not already fixed) in a relatively
short period of time.
A much more fundamental problem for commercial software vendors is the
lack of a central point of contact for communication and that is
partly solved by the existence of the commercial Linux vendors.
-Arun
Bob Withers
In article <01bd0b13$3da5c480$2e5c0c26@sfinance3>, s...@mindspring.com
says...
>
>
> Bjorn Borud <bo...@guardian.no> wrote in article
> <m2wwh3x5...@lucifer.guardian.no>...
> >
> > | <sigh> isn't this covered in the Linux FAQ? IF you can get hold of
> > | certain SCO binaries you can install Oracle just fine. I have done it
> on
> > | Red Hat and even on Slackware, both with the iBCS package installed of
> > | course.
> >
> > Linux and quite a few sources say that most large companies or
> > companies that have anything to do with the Internet, use Linux -- yet
> > Oracle are unable to deliver a Linux version of their database.
>
> Linux is missing many features for any OS to be worth considering
> as a viable database platform.
> 2. Linux has no transaction-oriented filesystem.
> 3. Linux has no support for raw devices - hence NO even remote possibility
> to run Oracle Parallel Server.
> laugh and puke steam.
> cannot do
> record-locking at all, last time I checked. So Oracle would have to
> write its own
> code for that. Why would anyone sane want that?
>
> almost
> from scratch.
I notice they managed to get an Oracle database running on 16 bit Windows
and Netware without writing their own version of those platforms.
S V
D. Jeff Dionne <je...@maribor.pfnet.com> wrote in article
<34982AEE...@maribor.pfnet.com>...
> [chop]
> > > strange really; it has become a FAQ how to install SCO binaries on
> > > Linux and quite a few sources say that most large companies or
> > > companies that have anything to do with the Internet, use Linux --
> > > yet
> > > Oracle are unable to deliver a Linux version of their database.
> >
> > Linux is missing many features for any OS to be worth considering
> > as a viable database platform.
>
> Yet it runs just fine under iBCS emulation on Linux...
"fine" in your terminology probably means "just being able to pop it up".
And in any case we discuss the native Oracle port, not the one
which runs under SCO libraries.
> > 1. Linux has no logical volumes layer.
> > 2. Linux has no transaction-oriented filesystem.
>
> So write support for these, however...
> It runs just fine under iBCS emulation on Linux...
see above
> > 3. Linux has no support for raw devices - hence NO even remote
> > possibility
> > to run Oracle Parallel Server.
>
> Have no idea what you mean. raw devices work just fine on Linux,
> tar -xvf /dev/fd0 for instance? Works fine.
you have no clue what "raw device" means. /dev/fd0 is block device.
Oracle needs character devices for disk partitions.
Do you have in Linux something like /dev/rhda1 (character device)
in addition to /dev/hda1(block device?). NO.
So go buy a clue, dude
> If Oracle needs raw
> partitions, no problem.
Puke
> > 4. Linux networking is flaky at best.
>
> Then the whole internet is flakey, since Linux runs on more servers
> on the net than anything else. Rock solid is more like it.
> It runs just fine under iBCS emulation on Linux...
>
> > 5. Linux OS block size is what? 512 bytes? It would make even MS
> > Access laugh and puke steam.
>
> You're thinking of (an old version of) SunOS :-)
>
> > 6. Linux SMP is rudimentary and flaky at best.
>
> It's actually not that bad in 2.1 I hear, and I fail to see how that
> affects the viability of Oracle on Linux.
>
> > 7. The same for multithreading.
>
> But it works fine under iBCS on Linux. Interesting. You're mistaken.
>
> > 8. Oracle requires efficient IPC and record-locking mechanisms. Linux
> > cannot do
> > record-locking at all, last time I checked. So Oracle would have
to
> > write its own
> > code for that. Why would anyone sane want that?
>
> But it works fine under iBCS on Linux. Interesting. You're mistaken
>
> >
> > In fact porting Oracle to Linux would require Oracle to write its own
Linux
> > almost
> > from scratch.
>
Nix
In article <01bd0b13$3da5c480$2e5c0c26@sfinance3>,
S V <s...@mindspring.com> wrote:
>
> Linux is missing many features for any OS to be worth considering
> as a viable database platform.
What's the VFS, then?
(Assuming that that's what you mean by a `logical volumes layer'. Linux's
fs, like Unix's fs, doesn't volume-ise its directory structure - and I
haven't seen anyone calling other Unixes, to which ORACLE has been
ported, a `nonviable database platform'.)
> 2. Linux has no transaction-oriented filesystem.
That's annoying. One is coming. There is a web page, but I've forgotten
where.
> 3. Linux has no support for raw devices - hence NO even remote possibility
> to run Oracle Parallel Server.
!!! Complete and utter bollocks! What do you think /dev/hd?? are?
> 4. Linux networking is flaky at best.
Evidence? ISPs are using it as a networking platform; companies are using it
as a network gateway - hardly the sign of a `flaky' networking subsystem.
You must be trolling.
> 5. Linux OS block size is what? 512 bytes? It would make even MS Access
> laugh and puke steam.
Depends on the fs. ext2fs has a 1Kb block size+512bytes for the inode.
Not that I can see how the block size affects a db at all; ORACLE, for one,
uses single big files as its datafiles and very rarely resizes them. And
Linux has pretty damn good disk caching.
> 6. Linux SMP is rudimentary and flaky at best.
I don't know this area well, but have heard the same. However, given that
Linus has multi-processor Alphas &c of his own now I imagine this area is
getting rapidly better :)
(Besides, it *can't* be as bad as NT's SMP `support'!)
> 7. The same for multithreading.
libc5 had very poor multithreading support, but glibc6 is fine. The *OS*'s
support, however, has been fine for most of 2.0.x and all of 2.1.x; and a
portable db should have a virtualisation layer wrapped around all external
libraries in any case. (All my portable code does.)
Threads aren't a built in part of the OS, but there is POSIX thread support,
and clone(2) lets that work OK. It's not a full implementation, yet.
> 8. Oracle requires efficient IPC and record-locking mechanisms. Linux
> cannot do
> record-locking at all, last time I checked. So Oracle would have to
> write its own
> code for that. Why would anyone sane want that?
Er, as you just said, ORACLE Parallel Server hits the disk directly,
rendering questions of what the OS's filesystem-level access code does
completely irrelevant. Oracle *have* written their own locking code. Besides,
when you put the load of proper parallelisation & multiuser synch code in
there, the programming time load of writing simple physical-level record
locking code is absolutely negligible.
> In fact porting Oracle to Linux would require Oracle to write its own Linux
> almost from scratch.
Complete and utter tosh, like most of your post.
> ==> I don't see much (or rather _any) prospects for seeing Oracle on
> Linux.
<politics mode=rms>
I don't see any because Oracle is a big corporation and big corporations
(with the notable exception of DEC :} ) seem to think atm that if they ignore
this non-capitalist free software idea, it will go away.
</politics>
--
"Computers in the future may weigh no more than 1.5 tons."
-- Popular Mechanics, forecasting the relentless march of
science, 1949
o r c e l l . p o r t l a n d . o r . u s
><m2wwh3x5...@lucifer.guardian.no>...
>>
>> | <sigh> isn't this covered in the Linux FAQ? IF you can get hold of
>> | certain SCO binaries you can install Oracle just fine. I have done it
>on
>> | Red Hat and even on Slackware, both with the iBCS package installed of
>> | course.
>>
>> Linux and quite a few sources say that most large companies or
>> companies that have anything to do with the Internet, use Linux -- yet
>> Oracle are unable to deliver a Linux version of their database.
>
>as a viable database platform.
>1. Linux has no logical volumes layer.
You don't need that for a database.
>2. Linux has no transaction-oriented filesystem.
You don't care about that for a database.
>3. Linux has no support for raw devices - hence NO even remote possibility
> to run Oracle Parallel Server.
The only way this becomes an issue is if the system buffering gets in
the way of the database buffering, since any serious high-reliability
database would want to talk directly to the disk.
>4. Linux networking is flaky at best.
News to me.
>5. Linux OS block size is what? 512 bytes? It would make even MS Access
>laugh and puke steam.
The _disk_ block size is 512 bytes, which, not surprisingly, is the
case for almost every operating system under the sun, since that's
how you get disks.
>6. Linux SMP is rudimentary and flaky at best.
Depends on the version (and you certainly don't need multiple processors
to run a relational database.)
>7. The same for multithreading.
You're wrong here too.
>8. Oracle requires efficient IPC and record-locking mechanisms. Linux
>cannot do
> record-locking at all, last time I checked. So Oracle would have to
>write its own
> code for that. Why would anyone sane want that?
If you're talking to a raw disk, you need to do your own locking.
Linux's IPC is good enough for the job.
>In fact porting Oracle to Linux would require Oracle to write its own Linux
>almost
>from scratch.
Nope. Just a raw device layer, at most. Oh, and customers who are
willing to pay for it.
____
david parsons \bi/ Oracle runs on NT, doesn't it?
\/
Josue Emmanuel Amaro
In article <slrn69h9a4....@knuth.brownes.org>, cbbr...@hex.net wrote:
>
> Oracle doesn't have a Linux version because they don't care to have a Linux
> version. There aren't enough suited people in the Linux community to
> "schmooze," and that's what Oracle is *really* into. Oracle is a great
> sales organization that happens to have a fairly decent database product.
> Much as IBM was, in the '60s and '70s, the most powerful *sales*
> organization on Planet Earth.
Probably Oracle does not provide a Linux version because none of its
current paying customers have made a fuss about it. For the same reason
that there isn't a version of Oracle for Windows95 or 3.1. ( There is
lite version, but we are talking server ( big database ) here. )
There may be a business for Oracle to do a Linux version.
Pro: They can sell the database, with an addition of a Free OS.
Con: Unfortunatelly it would be kind of hard since they would have create
a "Linux" support division since they would not only have to support their
product, but also provide support for the operating system.
But the reality is that no one has approached Oracle with a requirement (
and maybe some development money. )
Lets face it, management at Oracle will simple answer with "show me the money".
Believe me, if someone can come up with a business plan with some hard
numbers, maybe someone will listen.
> --
> Christopher B. Browne, cbbr...@hex.net, chris_...@sdt.com
Josue
Christopher B. Browne
On Wed, 17 Dec 1997 14:41:34 -0500, D. Jeff Dionne <je...@maribor.pfnet.com>
posted:
>[chop]
>> > strange really; it has become a FAQ how to install SCO binaries on
>> > Linux and quite a few sources say that most large companies or
>> > companies that have anything to do with the Internet, use Linux --
>> > yet
>> > Oracle are unable to deliver a Linux version of their database.
>>
>> Linux is missing many features for any OS to be worth considering
>> as a viable database platform.
>
>
>
>It runs just fine under iBCS emulation on Linux...
Both are under construction...
>> 3. Linux has no support for raw devices - hence NO even remote
>> possibility
>> to run Oracle Parallel Server.
>
>Have no idea what you mean. raw devices work just fine on Linux,
>tar -xvf /dev/fd0 for instance? Works fine. If Oracle needs raw
>partitions, no problem.
Raw devices are getting to be less and less of an issue over time what with
the increasing levels of buffering being provided at the hardware level.
Linux does *not* at this time provide "raw" devices in the sense that Oracle
(and I believe Informix) expects; it provides "cooked" devices where
cacheing and buffering is *guaranteed* to take place at the OS level.
Oracle Parallel Server is probably not the first product that would be
likely to be ported to UNIX, and I've never worked with an Oracle
installation yet that didn't use "cooked" disks, so the point is moot for a
lot of even rather large installations...
>> 4. Linux networking is flaky at best.
>
>Then the whole internet is flakey, since Linux runs on more servers
>on the net than anything else. Rock solid is more like it.
>It runs just fine under iBCS emulation on Linux...
The BSDs have a longer history of high levels of stability, and probably
still maintain superiority over Linux. The issue being brought up is not
whether Linux has "reasonably stable" networking, but rather whether it is
robust under extreme conditions. Linux is a whole lot more stable than it
used to be; certainly moreso than some of the platforms that Oracle runs on
(NT, anyone?), but there is still room for improvement.
>> 5. Linux OS block size is what? 512 bytes? It would make even MS
>> Access laugh and puke steam.
>
>You're thinking of (an old version of) SunOS :-)
Variable, configurable block sizes have been around for quite a long time
now. The claim that block sizes are limited to 512 bytes, and that this
would necessarily be a problem, is certainly a silly one.
>> 6. Linux SMP is rudimentary and flaky at best.
>
>It's actually not that bad in 2.1 I hear, and I fail to see how that
>affects the viability of Oracle on Linux.
It's a fair enough criticism for those that would want to build database
servers with 1GB of RAM, 300GB of disk, and multiple SCSI controllers to
handle huge loads. You probably want to have 4 processors on that kind of
box...
>> 7. The same for multithreading.
>
>But it works fine under iBCS on Linux. Interesting. You're mistaken.
Give threading six months to solidify under the GLIBC "regime," and we can
provide a somewhat different story...
>> 8. Oracle requires efficient IPC and record-locking mechanisms. Linux
>> cannot do
>> record-locking at all, last time I checked. So Oracle would have to
>> write its own
>> code for that. Why would anyone sane want that?
>
>But it works fine under iBCS on Linux. Interesting. You're mistaken.
It's more interesting to consider whether or not this is something that
Oracle normally needs to implement on the various systems that it supports.
Sybase had similar concerns; they then lost their heads over the issue that
if they patched the kernel to provide the kind of locking semantics that
they wanted, they would need to GPL the kernel patch, and they didn't want
that.
It's fairly unlikely that there are any changes required here that could not
be readily handled if the Linux community were provided some sort of
specification as to what the required locking semantics would be.
>> In fact porting Oracle to Linux would require Oracle to write its own Linux
>> almost
>> from scratch.
>
>But it works fine under iBCS on Linux. Interesting. You're mistaken
People run "toy" installations of SCO/Oracle atop Linux. SCO is not one of
the high volume sorts of platforms; as a result, it's unlikely that anyone
is throwing the kinds of challenging database "traffic" at the servers that
would tickle the problems that would make them choke.
>> ==> I don't see much (or rather _any) prospects for seeing Oracle on
>> Linux.
>
>That may be so, but it has nothing to do with weakness in Linux. I
>suspect it has to do with Oracle's relationship with the Big UN*X
>Vendors.
There *are* some weaknesses. If Oracle (or Sybase or Informix) were to
suggest a list of the most critical things that would need to be patched to
get Linux "up to scratch" as a supportable platform, and the above list of 8
items overstates what would realistically be needed, there's nothing that
couldn't be fairly quickly provided.
Oracle doesn't have a Linux version because they don't care to have a Linux
version. There aren't enough suited people in the Linux community to
"schmooze," and that's what Oracle is *really* into. Oracle is a great
sales organization that happens to have a fairly decent database product.
Much as IBM was, in the '60s and '70s, the most powerful *sales*
organization on Planet Earth. (The fact of selling computers being largely
incidental.) Microsoft has the greatest legal manipulation and acquisitions
group of any company in modern times, as well as some sharp marketers.
(Informix, on the other hand, has what seems to be a better database
product, but they seem to be having trouble selling their way out of a wet
paper bag.)
How much any of this has to do with whether Linux does or does not have any
particular features is pretty hard to grasp at.
--
Christopher B. Browne, cbbr...@hex.net, chris_...@sdt.com
PGP Fingerprint: 10 5A 20 3C 39 5A D3 12 D9 54 26 22 FF 1F E9 16
URL: <http://www.hex.net/~cbbrowne/>
Bill Gates to his broker: "You idiot, I said $150 million on **SNAPPLE**!!!"
Matthew Hannigan
In article <679vcl$5...@pell.pell.portland.or.us>,
david parsons <o r c @ p e l l . p o r t l a n d . o r . u s> wrote:
>In article <01bd0b13$3da5c480$2e5c0c26@sfinance3>,
>S V <s...@mindspring.com> wrote:
>>as a viable database platform.
>
>
>
Yeah, last I heard the db does transactions, not the fs!
>>3. Linux has no support for raw devices - hence NO even remote possibility
>> to run Oracle Parallel Server.
>
> The only way this becomes an issue is if the system buffering gets in
> the way of the database buffering, since any serious high-reliability
> database would want to talk directly to the disk.
In official Oracle documentation it says that using raw partitions
can in some circumstances _reduce_ performance, because the OS does
a better job of buffering. It is a lot less flexible anyway.
For high-reliablity, can't you tell ext2fs to do synchronous writes now?
>> [other crap deleted]
What's the bet that someone on the Oracle development team _has_
actually ported Oracle to Linux, even if it just for his/her own
amusement? (people from Oracle, this is your cue to make an
anonymous posting :-)
PS. David, it's polite to point out that you've changed follow-ups.
PPS. Note follow-up to:
--
-Matt
Mark Hahn
: Linux is missing many features for any OS to be worth considering
: as a viable database platform.
this author is uninformed.
: 1. Linux has no logical volumes layer.
MD exists, though it's very close to a full LVM.
: 2. Linux has no transaction-oriented filesystem.
thank god! now, please explain how you want a FS to be ACID, and why
this is relevant to a database.
: 3. Linux has no support for raw devices - hence NO even remote possibility
: to run Oracle Parallel Server.
DB people are acolytes of a black religion. one of their main items of
faith is that you must have raw devices. as such, there's no reason to
ask for a defense of this item. besides, didn't you want an ACID FS,
rather than raw devices? also, ever heard of O_SYNC?
: 4. Linux networking is flaky at best.
BULLSHIT. let me guess, you're a *BSD dweeb, right?
: 5. Linux OS block size is what? 512 bytes? It would make even MS Access
: laugh and puke steam.
no, asshole.
: 6. Linux SMP is rudimentary and flaky at best.
no, asshole.
: 7. The same for multithreading.
no, asshole.
: 8. Oracle requires efficient IPC and record-locking mechanisms. Linux
: cannot do
: record-locking at all, last time I checked. So Oracle would have to
: write its own
: code for that. Why would anyone sane want that?
actually, most serious transaction and DB systems _do_ definitely want
to do their own locking.
: Cordially,
yeah whatever.
summary: Linux is free and extremely fast/efficient on dirt-cheap hardware.
it's consciously aimed at _workstation_ performance, but a single competent
programmer could turn it into an excellent server platform in a month or two.
Brian Wheeler
In article <01bd0b13$3da5c480$2e5c0c26@sfinance3>,
"S V" <s...@mindspring.com> writes:
>
>
> Bjorn Borud <bo...@guardian.no> wrote in article
> <m2wwh3x5...@lucifer.guardian.no>...
>>
>> | <sigh> isn't this covered in the Linux FAQ? IF you can get hold of
>> | certain SCO binaries you can install Oracle just fine. I have done it
> on
>> | Red Hat and even on Slackware, both with the iBCS package installed of
>> | course.
>>
>> Linux and quite a few sources say that most large companies or
>> companies that have anything to do with the Internet, use Linux -- yet
>> Oracle are unable to deliver a Linux version of their database.
>
> as a viable database platform.
to answer from that perspective, since Serge is suggesting that linux isn't
suitable for *any* db vendor to consider porting to.
> 1. Linux has no logical volumes layer.
Why is this required? I use sybase on multiple platforms. Some have
logical volume *capability* some do not. It is __NOT REQUIRED__ for a RDBMS
to use the OSes logical volumes. Strike 1.
> 2. Linux has no transaction-oriented filesystem.
RDBMSes do not require a transaction-oriented filesystem...even when
the data is _STORED_ on the filesystem. In fact, if the data is stored on a
transaction-oriented filesystem, it often slows the RDBMS down. Strike 2.
> 3. Linux has no support for raw devices - hence NO even remote possibility
> to run Oracle Parallel Server.
Most RDBMSes that I've seen allow filesystem devices. In addition,
while not _RAW_ devices, per se, the /dev/?da? devices are virtually equivalent
in functionality...with the exception that the linux devices have buffering and
proper raw devices do not. All this means is that your data is not guaranteed
to make it to the disk when the RDBMS thinks it is. Not a problem for most
users, and those who require this probably wouldn't be running linux anyway.
Strike 3.
> 4. Linux networking is flaky at best.
That's odd. On Suns, its actually faster than Sun's own OS. You
really don't know what you're talking about. In any case, networking speed
is not a critical factor for RDBMS performance. Strike 4.
> 5. Linux OS block size is what? 512 bytes? It would make even MS Access
> laugh and puke steam.
Block size is dependant on the device in question. You have proved
yourself a troll. Come to think of it, I've seen you troll before. Strike 5.
> 6. Linux SMP is rudimentary and flaky at best.
While this is _partially true_ (Linux SMP IS rudimentary...in 2.0), it
is by no means flaky. How many people run RDBMSes on an SMP box anyway?
> 7. The same for multithreading.
You should really KNOW WHAT YOU'RE TALKING ABOUT BEFORE YOU WRITE.
Of course, as a troll, you probably don't care that Linux multithreading is
just fine.
> 8. Oracle requires efficient IPC and record-locking mechanisms. Linux
> cannot do
> record-locking at all, last time I checked. So Oracle would have to
> write its own
> code for that. Why would anyone sane want that?
IPC under linux is quite efficient. Record locking happens within the
RDBMS _REGARDLESS OF PLATFORM_. Do you have any clue what you are typing, or
are you so pro NT (I've seen you troll before), that you must try to make all
other OSes look bad?
>
> In fact porting Oracle to Linux would require Oracle to write its own Linux
> almost
> from scratch.
>
> ==> I don't see much (or rather _any) prospects for seeing Oracle on
> Linux.
>
> Cordially,
> Serge
It must hurt to be that desperate for attention that you have to troll.
Brian
Brian Wheeler
In article <01bd0b3a$6cfa65c0$66601b26@sfinance3>,
"S V" <s...@mindspring.com> writes:
>
>
> D. Jeff Dionne <je...@maribor.pfnet.com> wrote in article
> <34982AEE...@maribor.pfnet.com>...
>> [chop]
>> > > Linux and quite a few sources say that most large companies or
>> > > companies that have anything to do with the Internet, use Linux --
>> > > yet
>> > > Oracle are unable to deliver a Linux version of their database.
>> >
>> > Linux is missing many features for any OS to be worth considering
>> > as a viable database platform.
>>
>
>
> And in any case we discuss the native Oracle port, not the one
> which runs under SCO libraries.
Since you obviously don't know what you're talking about, I'll give
you a clue:
iBCS is a way to run other OS binaries on Linux. The *LINUX* system
calls are used, instead of the _real_ OS's system calls. There's really not
much emulation to it. In any case, since linux is really doing all the work
(the libraries are just calling Linux instead of the OS in question),
LINUX IS CAPABLE OF RUNNING ORACLE. In fact, it has done so. That is not
to say that a native port of it wouldn't be a good thing (there is some
overhead to iBCS) but if the SCO version of Oracle works ok on the Linux
kernel, then all the features available to Oracle in SCO should also be
available to Oracle on Linux.
>
>> > 1. Linux has no logical volumes layer.
>> > 2. Linux has no transaction-oriented filesystem.
>>
>> So write support for these, however...
>> It runs just fine under iBCS emulation on Linux...
>
> see above
>
>
>> > 3. Linux has no support for raw devices - hence NO even remote
>> > possibility
>> > to run Oracle Parallel Server.
>>
>> Have no idea what you mean. raw devices work just fine on Linux,
>> tar -xvf /dev/fd0 for instance? Works fine.
>
> you have no clue what "raw device" means. /dev/fd0 is block device.
> Oracle needs character devices for disk partitions.
> Do you have in Linux something like /dev/rhda1 (character device)
> in addition to /dev/hda1(block device?). NO.
> So go buy a clue, dude
The only real difference between /dev/hda1 and /dev/rhda1 (as you put
it), is the fact hda1 is block and buffered, and a proper raw device (you're
calling rhda1) is a char device and unbuffered. Ever heard of a filesystem
device? I assume Oracle supports them. Sybase does. No raw device really
needed.
Brian Wheeler
bdwh...@indiana.edu
Bjorn Borud
["S V" <s...@mindspring.com>]
keep in mind that those who would want to run Oracle on Linux usually
have pretty modest database needs, thus much of what you're asking for
isn't really an issue at all. if you need a large SMP system or
terabytes of tablespaces you would probably buy something else than a
PC -- don't you agree?
| Linux is missing many features for any OS to be worth considering
| as a viable database platform.
| 1. Linux has no logical volumes layer.
what's the problem? I have worked on ONE site where they used logical
volumes, but they would have done just as well without. this is a non
issue.
| 2. Linux has no transaction-oriented filesystem.
would you like to explain why you think this is an issue at all?
because I really can't see any reason why you would need one. sure,
the extra redundancy at the file system level is probably a cool
feature, but this has nothing to do with the database.
| 3. Linux has no support for raw devices - hence NO even remote possibility
| to run Oracle Parallel Server.
I have yet to desire raw devices for using Oracle and I certainly have
never needed to run Oracle Parallel server. if I would ever need to
run it I would certainly not do it on a PC.
| 4. Linux networking is flaky at best.
uh, compared to what? would you mind explaining to me how you got
that impression, because we're obviously from different worlds. in my
world Linux holds up pretty good compared to other OSes running on PCs
and more traditional UNIX servers as well.
| 5. Linux OS block size is what? 512 bytes? It would make even MS
| Access laugh and puke steam.
uh, what block size would that be? if you're talking about the file
system one can only guess whatever gave you that idea. on my system
the block size on a ext2 filesystem can be between 1024 and 4096.
would you care to explain what block size you are referring to and
what block sizes other systems where Oracle is available use?
| 6. Linux SMP is rudimentary and flaky at best.
that might be true, but still, they ported Oracle to NT and last time
I checked Linux scaled better across 2 CPUs than NT.
would you care to explain?
| 7. The same for multithreading.
from what I understand Linux multithreading is different from for
example Solaris. I'm told (by people who are intimately familiar
with the Linux kernel) that Linux does multithreading faster than for
example Solaris.
it's there, it's fast -- now, would you care to explain exactly why
you think that Linux has "rudimentary and flaky at best"
multithreading?
| In fact porting Oracle to Linux would require Oracle to write its
| own Linux almost from scratch.
you are most absolutely wrong.
| ==> I don't see much (or rather _any) prospects for seeing Oracle on
| Linux.
perhaps it's time for you to take a closer look at Linux as well as
other systems that Oracle has been ported to before you jump to any
conclusions about the suitability of Linux as a platform for running
Oracle. Oracle has been ported to operating systems that are _far_
less advanced than Linux and that lack almost everything that you
mentioned in your post. that hasn't stopped Oracle from porting to
them so far, so you will have to think of something better.
the fact that Oracle already can be run on Linux using iBCS2 (this has
been possible since at least version 1.2 of the Linux kernel) should
prove that Oracle can in fact run under Linux without, as you state,
Oracle Corp. having to rewrite Linux from scratch.
(Some time ago I read that someone had benchmarked Oracle on a PC using
both SCO UNIX and Linux with some very interesting results. does
anyone remember who did this?)
porting Oracle to Linux is certainly not a technical issue, it is a
political and economical one. if the CEO of Oracle is as dedicated to
the cause of fighting Microsofts operating system monopoly as he would
like us to think he is it puts him in a strange light when he doesn't
pay attention to those who use one of the more viable alternatives
to NT turn to Oracle for support.
-Bjørn
Peter Schwing
In article <01bd0b13$3da5c480$2e5c0c26@sfinance3>, "S V" <s...@mindspring.com> writes:
> ...
> Linux is missing many features for any OS to be worth considering
> as a viable database platform.
> 1. Linux has no logical volumes layer.
> to run Oracle Parallel Server.
> laugh and puke steam.
> cannot do
> record-locking at all, last time I checked. So Oracle would have to
> write its own
> code for that. Why would anyone sane want that?
>
> almost
> from scratch.
>
> Linux.
>
> Serge
>
Have you ever Adabas D for Linux. It has raw-device support. And the other
points you mentioned didn't seem to stop a port of Adabas D to Linux.
Peter
Kristian Koehntopp
"S V" <s...@mindspring.com> writes:
>Linux is missing many features for any OS to be worth considering
>as a viable database platform.
Oracle already _is_ running on Linux, via the iBCS system call
converter (I would not want to call that 'emulation', because as
an emulator, iBCS has not that much to do).
The thing that is being called for is a native version of Oracle
on Linux. And it is not being called for because of speed (iBCS
"emulation" costs not much speed, because of its simplicity),
but because of better support and easier installation.
>1. Linux has no logical volumes layer.
Linux ha the md driver, which does build RAID 0, RAID 1, RAID
0+1 and RAID 5 arrays from a number of physical drives.
>2. Linux has no transaction-oriented filesystem.
True. But Oracle does not need a transaction oriented
filesystem, because it does implement transactions and logs
itself, as does any database worth its money.
>3. Linux has no support for raw devices - hence NO even remote possibility
> to run Oracle Parallel Server.
Linux has support for raw devices, but not for character special
raw devices. That is not really a problem, because there is
really no difference in the access method from the viewpoint of
a program. The difference between character special raw devices
and block special raw devices is the buffer cache (which is NOT
being used with character special devices).
Oracle wants to use character special (unbuffered) devices to
make sure that all data is on disk and not in a buffer cache
when a transaction is complete. That can be had with Linux block
special devices as well (just flush them).
>4. Linux networking is flaky at best.
That has been true for the very first Linux networking stack.
Linux is at it's third revision of networking code now and the
Linux networking stack is in the best-of-class league now. Linux
networking implements many performance enhancements and
redesigns that Van Jacobson suggested for the 4.3BSD networking
stack networking and Linux networking is fastest over the wire
and even outperforms Sun. Go to Dejanews ad seek out the
discussion Alan Cox had with the BSD bunch over Linux NET3 vs.
BSD networking and you will learn some very interesting facts.
>5. Linux OS block size is what? 512 bytes? It would make even MS Access
>laugh and puke steam.
The ext2 file system as a block size of either 1024, 2048 or
4096 byte per block. But that is only relevant, if you are going
to put your database into dbf-files. If you do that, you are not
using raw devices (see your point 3).
>6. Linux SMP is rudimentary and flaky at best.
I am writing this answer from a Dual Pentium 200 (Asus Board).
This machine is useable and stable. I am using Linux 2.0.32,
because I found that to be faster and more stable than Windows
NT 4.0 SP3 on the same hardware.
>7. The same for multithreading.
Linux has the clone() system call, which is a very beautiful and
clever generalization of fork(). On top of clone() exists the
linuxthreads library, which uses clone() to implement kernel
level threads. Both clone() and linuxthreads are quite stable
and very useable.
>8. Oracle requires efficient IPC and record-locking mechanisms. Linux
> cannot do record-locking at all, last time I checked.
You should check again to bring your linux knowledge up to date.
In this particular case you should have a look at
/usr/src/linux/fs/locks.c, which contains the code for the
fcntl() and lockf() system calls (Posix compatible locks) and
for the flock() system call (which has a different locking
personality and semantic).
Linux does have deficiencies and limitations, for example with
very large files and very large filesystems. These limitations
are being worked on at the moment with the 2.1.x development
kernels and will be publicly available in the 2.2.x stable
kernel.
Until that these limitations can be worked around by using raw
devices instead of dbf-files in a filesystem or by splitting the
database into multiple dbf-files. Actually, the latter is
recommended for certain setups even _if_ your filesystem can
handle larger files.
Kristian
--
Kristian Koehntopp, Wassilystrasse 30, 24113 Kiel, +49 431 688897
http://www.koehntopp.de
"Bitte zwei Zeilen Rangierabstand lassen."
Kristian Koehntopp
cbbr...@news.brownes.org (Christopher B. Browne) writes:
>Oracle doesn't have a Linux version because they don't care to
>have a Linux version.
And _this_ is the critical point in your argumentation.
From my personal experience with Oracle, "don't care" is the
perfect description of the attitude of Oracle, the company.
Working in a bank for a living, I have _never_ seen more
arrogant "bring us to your leaders, we don't talk to you
earthling" sales droids invading our beloved little IT
department. I was not the only one getting that impression that
day, so our strategic company database product is now not
Oracle...
But that is a different story to be told another time,
Nilanjan Bhowmik
Josue Emmanuel Amaro wrote:
>
>
> Lets face it, management at Oracle will simple answer with "show me the money".
>
> Believe me, if someone can come up with a business plan with some hard
> numbers, maybe someone will listen.
>
I think this is the most sensible idea anyone ever posted in this
thread. Is there any Linux Geek converted to Business Grad available
for this small project.
If we can create a solid business plan and may be small presentation
( anyone here has pull in ORACLE ? ) to them I think they will consider
it. We may need to involve Caldera in it - they may be trying it by now
but it does not harm to give them support for it.
I can personally start asking this questions to the people I know
if that does any good.
Anyone care to join.
Nilanjan
Robert M. Stockmann
Bob Withers wrote:
>
> In article <01bd0b13$3da5c480$2e5c0c26@sfinance3>, s...@mindspring.com
> says...
> >
> >
> > Bjorn Borud <bo...@guardian.no> wrote in article
> > <m2wwh3x5...@lucifer.guardian.no>...
> > >
> > > | <sigh> isn't this covered in the Linux FAQ? IF you can get hold of
> > > | certain SCO binaries you can install Oracle just fine. I have done it
> > on
> > > | Red Hat and even on Slackware, both with the iBCS package installed of
> > > | course.
> > >
> > > Linux and quite a few sources say that most large companies or
> > > companies that have anything to do with the Internet, use Linux -- yet
> > > Oracle are unable to deliver a Linux version of their database.
> >
> > as a viable database platform.
> > to run Oracle Parallel Server.
> > laugh and puke steam.
> > cannot do
> > write its own
> > code for that. Why would anyone sane want that?
> >
> > In fact porting Oracle to Linux would require Oracle to write its own Linux
> > almost
> > from scratch.
>
> and Netware without writing their own version of those platforms.
Well this is the best argument I have seen in this thread!
Oracle can run native on Linux, of course it can!
Robert
--
++---------------------------++----------------------------------------++
|| R.M. Stockmann || InfoMagic Nederland VOF
||
|| st...@infomagic.nl || Unix administration & support
||
|| http://www.infomagic.nl || The Netherlands ||
++---------------------------++----------------------------------------++
250-Linux: A copylefted Unix-like operating system for 80[3456]86,
250- DEC Alpha, Sun SPARC, Sun UltraSPARC, Motorola 68k,
250- PowerPC/PowerMac, ARM, Mips R[3,4]x00, Fujitsu AP/1000+
250- and more to come.
Christopher B. Browne
On 17 Dec 1997 11:14:21 -0800, Arun Sharma <ash...@this.is.not.me.netscape.com> posted:
>
>> 1. Linux has no logical volumes layer.
>> 2. Linux has no transaction-oriented filesystem.
>> 3. Linux has no support for raw devices - hence NO even remote possibility
>> to run Oracle Parallel Server.
>
>validity of your data.
As has been nicely discussed elsewhere, 2. and 3. are mutually contradictory
claims. If it is important to have a transaction-oriented filesystem, then
Oracle is *obviously* not using a raw device for data storage, and vice
versa.
Obvious trolling.
>> 4. Linux networking is flaky at best.
>
>And how did you reach that conclusion ?
*Clear* trolling.
>> 5. Linux OS block size is what? 512 bytes? It would make even MS Access
>> laugh and puke steam.
>Pure misinformation. The block size is a file system attribute and can
>be changed.
Obvious trolling; I'm not sure how long it has been since the native block
size became *configurable* to up to about 4K; it's been some years now.
>> 6. Linux SMP is rudimentary and flaky at best.
>
>And what percentage of database servers run on multiprocessor boxes ?
This is probably the least bad criticism of the bunch. The bigger DB
servers *are* running on SMP boxes; Linux's support of SMP has been
improving from the state of "nonexistant" to being (at present) "seems
fairly good." It might soon be "up to scratch" for the purpose of supporting
a big database system. I wouldn't feel 100% comfortable at this stage.
>> 7. The same for multithreading.
>
>Misinformation again. Kernel level multithreading works perfectly fine
>with Linuxthreads - a POSIX threads implementation for Linux.
There are multiple implementations of threading to cover the different
possible models of threading; there certainly are libraries; the entrance of
GLIBC which actively supports threading should firm this area up
substantially over the next year.
>> 8. Oracle requires efficient IPC
>
>Ever checked the lmbench numbers for Linux Vs <your favorite OS here> ?
>
>> and record-locking mechanisms. Linux cannot do
>> record-locking at all, last time I checked. So Oracle
>> would have to write its own code for that. Why would anyone
>> sane want that?
>
>Not reason enough. If this is what is holding up Oracle, Linux
>developers can fix it (if it's not already fixed) in a relatively
>short period of time.
And this is frankly the most stupid "troll" of all the ones that could come
up. DBMS systems manage locking *themselves.* The R/3 application (one of
the *really* big database applications out there) adds its *own* locking
system, *ignoring* the multiplicity of mutually incompatible locking systems
implemented by the various DBMS vendors.
Record locking isn't something that a competent database designer would want
to entrust to the vagaries of "lock" implementations by the fifty-odd
operating system platforms supported by Oracle.
--
Christopher B. Browne, cbbr...@hex.net, chris_...@sdt.com
Tim Kutergin
S V (s...@mindspring.com) wrote:
: Bjorn Borud <bo...@guardian.no> wrote in article
: <m2wwh3x5...@lucifer.guardian.no>...
: Linux is missing many features for any OS to be worth considering
: as a viable database platform.
: 1. Linux has no logical volumes layer.
Multiple device driver is capable of merging several physical devices in
one logical.
BTW, SCO OpenServer 3, does not support logical volumes too. What the
difference ? Oracle really does not need LV, it can span databases
on several disks.
: 2. Linux has no transaction-oriented filesystem.
SCO OpenServer 3 has no such filesystem too. Oracle has not much need for
this type of filesystem, because it uses data files of fixed size, so
system crash does not cause file size problem. All other data protection
is provided by Oracle.
: 3. Linux has no support for raw devices - hence NO even remote possibility
: to run Oracle Parallel Server.
What is raw device ? I think this is device than could be operated with
read()/write() and may be seek(). Block device supports mount(). Device
drivers may support both interfaces, so one device file could provice
characted and block interfaces simultaneously.
Linux disk devices are capable of both operations, so why You need
separate device files? Don't mount it and feed it to Oracle. The real
problem is that Linux possibly caches disk devices with character interface,
that may be dangerous, because Oracle believes that anything written on
raw device is on disk already. But this could be fixed.
: 4. Linux networking is flaky at best.
Is not more flaky than on early versions of SCO
: 5. Linux OS block size is what? 512 bytes? It would make even MS Access
: laugh and puke steam.
I think that mkfs can specify block size. Moreover driver could read
several blocks at once, so there will be no or little performance penalty.
: 6. Linux SMP is rudimentary and flaky at best.
May be or may be not. Sorry, I have no experience with SMP on linux or
other OS.
: 7. The same for multithreading.
Oracle versions for Unix do not use kernel threads at all. Oracle runs
several processes that are communicating through shared memory. Dispatcher
processes are selecting one of free server processes to run query. It is
something like Apache Web Server "multithreading". IMHO this is more robust
than threads, because only some part of process memory is shared.
Therefore it is more likely to crash one of server processes than entire
server.
: 8. Oracle requires efficient IPC and record-locking mechanisms. Linux
: cannot do
: record-locking at all, last time I checked. So Oracle would have to
: write its own
: code for that. Why would anyone sane want that?
Not true. System V IPC package on linux seems working ok. Oracle really
needs shared memory or semafores from this package. Record locking on files
using fcntl() works also, but I am not sure that Oracle needs them. I think
that Oracle locks are quite different from record lock in files supported by
OS. I think Oracle locks are implemented as semafores or something in shared
memory.
: In fact porting Oracle to Linux would require Oracle to write its own Linux
: almost
: from scratch.
No No No. Linux is much like other Unixes so Oracle should not totally
rewrite (like for sucker's OS Windows NT). I think that Oracle is working
good on linux using IBCS2 becouse of it does not require something
extraordinary from OS. Oracle IS GREAT !
: ==> I don't see much (or rather _any) prospects for seeing Oracle on
: Linux.
May be, but for completely other reasons. Oracle products are
costly. Linux is free. If somebody uses Linux because it is free, will it pay
many $$$ for Oracle ? I think Oracle thinks that they could not earn money
on linux port for this reason.
BTR
Timofey Kutergin
Bjorn Borud
["S V" <s...@mindspring.com>]
|
| And in any case we discuss the native Oracle port, not the one
| which runs under SCO libraries.
the iBCS2 interface merely remaps system calls and provides wrappers
to some system calls in order to provide the semantics of the OS it is
supposed to mimic.
the point is: Oracle can run under Linux so there is no technical
reason it shouldn't be ported. also I don't think that it would be
that much of a job either given that Oracle is written to be portable
and it has been ported to stranger OSes than Linux.
| Oracle needs character devices for disk partitions.
you don't really need raw devices; it's a feature you can live without
in most cases. if you need them you simply run Oracle on some other
platform. usually the OS will have the best idea about how the I/O
should be handled in order to get the best performance -- in fact I
have heard that some sites actually use Network Appliances servers
(instead of local file systems) in order to speed up their databases.
still, a native port to Linux would be a Good Thing in any case, since
that would guarantee the availability of client software. it is often
desirable to use an Oracle database from applications (like
web-servers) running on a Linux machine. today you can do that, but
not in any particularly elegant way.
anyhow, you don't seem to know much about Linux (or databases for that
matter) so I don't understand why you feel that you have to contribute
to the discussion. most of your assumptions are misinformed,
contradictory and sometimes completely irrelevant to the issue.
perhaps it would be better if you just admitted you don't really know
what you're talking about or just shut up.
Hannu Krosing
Bjorn Borud wrote:
> ["S V" <s...@mindspring.com>]
>
> keep in mind that those who would want to run Oracle on Linux usually
> have pretty modest database needs, thus much of what you're asking for
> isn't really an issue at all. if you need a large SMP system or
> terabytes of tablespaces you would probably buy something else than a
> PC -- don't you agree?
But you could still use linux ;)
> || 3. Linux has no support for raw devices - hence NO even remote possibility
> | to run Oracle Parallel Server.
>
> I have yet to desire raw devices for using Oracle and I certainly have
> never needed to run Oracle Parallel server. if I would ever need to
> run it I would certainly not do it on a PC.
What are the cooked devices then located on?
I think that SV simply does not know what a raw device is.
> | 5. Linux OS block size is what? 512 bytes? It would make even MS
> | Access laugh and puke steam.
>
> uh, what block size would that be? if you're talking about the file
> system one can only guess whatever gave you that idea. on my system
> the block size on a ext2 filesystem can be between 1024 and 4096.
>
> would you care to explain what block size you are referring to and
> what block sizes other systems where Oracle is available use?
I think he is referring to sector size, which is 512 for most unixen
> | ==> I don't see much (or rather _any) prospects for seeing Oracle on
> | Linux.
>
> perhaps it's time for you to take a closer look at Linux as well as
> other systems that Oracle has been ported to before you jump to any
> conclusions about the suitability of Linux as a platform for running
> Oracle. Oracle has been ported to operating systems that are _far_
> less advanced than Linux and that lack almost everything that you
> mentioned in your post. that hasn't stopped Oracle from porting to
> them so far, so you will have to think of something better.
It is ported even to a non-os, the Novell Netware server.
Hannu
Christopher Browne
On Fri, 19 Dec 1997 22:05:38 +0200, Hannu Krosing <ha...@sid.trust.ee> wrote:
>Bjorn Borud wrote:
>> || 3. Linux has no support for raw devices - hence NO even remote possibility
>> | to run Oracle Parallel Server.
>>
>> I have yet to desire raw devices for using Oracle and I certainly have
>> never needed to run Oracle Parallel server. if I would ever need to
>> run it I would certainly not do it on a PC.
>
>What are the cooked devices then located on?
>
>I think that SV simply does not know what a raw device is.
No, dome of the less knowledgeable Linux users don't know what a
*truly* raw device is, because Linux shields them from such.
A *truly* raw device will have *NO* kernel-based buffering or cacheing
of data.
--
"Bother," said Pooh, "Eeyore, ready two photon torpedoes and lock
phasers on the Heffalump, Piglet, meet me in transporter room three"
cbbr...@hex.net - "What have you contributed to Linux today?..."
<http://www.hex.net/~cbbrowne/lsf.html>
Joel Garry
In article <3499C383...@infomagic.nl>,
Robert M. Stockmann <st...@infomagic.nl> wrote:
>Bob Withers wrote:
>>
>> In article <01bd0b13$3da5c480$2e5c0c26@sfinance3>, s...@mindspring.com
>> says...
>> >
>> >
>> >
>> > almost
>> > from scratch.
>>
>> and Netware without writing their own version of those platforms.
>
>Well this is the best argument I have seen in this thread!
For fairly broad values of "running." :)
>
>Oracle can run native on Linux, of course it can!
>
>Robert
>--
>++---------------------------++----------------------------------------++
>|| R.M. Stockmann || InfoMagic Nederland VOF
>||
>|| st...@infomagic.nl || Unix administration & support
>||
>|| http://www.infomagic.nl || The Netherlands ||
>++---------------------------++----------------------------------------++
>250-Linux: A copylefted Unix-like operating system for 80[3456]86,
>250- DEC Alpha, Sun SPARC, Sun UltraSPARC, Motorola 68k,
>250- PowerPC/PowerMac, ARM, Mips R[3,4]x00, Fujitsu AP/1000+
>250- and more to come.
--
These opinions are my own and not necessarily those of Information Quest
jga...@eiq.com http://www.informationquest.com
http://ourworld.compuserve.com/homepages/joel_garry
"See your DBA?" I AM the @#%*& DBA!
Joel Garry
In article <m267omw...@lucifer.guardian.no>,
Bjorn Borud <bo...@guardian.no> wrote:
>["S V" <s...@mindspring.com>]
>
>keep in mind that those who would want to run Oracle on Linux usually
>have pretty modest database needs, thus much of what you're asking for
>isn't really an issue at all. if you need a large SMP system or
>terabytes of tablespaces you would probably buy something else than a
>PC -- don't you agree?
>
>| as a viable database platform.
>| 1. Linux has no logical volumes layer.
>
>volumes, but they would have done just as well without. this is a non
>issue.
Not a non-issue for places that use raw file-systems.
But I agree, they are not _necessary_.
>
>| 2. Linux has no transaction-oriented filesystem.
>
>would you like to explain why you think this is an issue at all?
>because I really can't see any reason why you would need one. sure,
>the extra redundancy at the file system level is probably a cool
>feature, but this has nothing to do with the database.
>
>| 3. Linux has no support for raw devices - hence NO even remote possibility
>| to run Oracle Parallel Server.
>
>I have yet to desire raw devices for using Oracle and I certainly have
>never needed to run Oracle Parallel server. if I would ever need to
>run it I would certainly not do it on a PC.
With the top-end MP PC's, PC is a misnomer. I don't think parallel
servers are a bad topic for this at all. Too bad the original post was
such a troll.
>
>| 4. Linux networking is flaky at best.
>
>uh, compared to what? would you mind explaining to me how you got
>that impression, because we're obviously from different worlds. in my
>world Linux holds up pretty good compared to other OSes running on PCs
>and more traditional UNIX servers as well.
>
>| 5. Linux OS block size is what? 512 bytes? It would make even MS
>| Access laugh and puke steam.
>
>uh, what block size would that be? if you're talking about the file
>system one can only guess whatever gave you that idea. on my system
>the block size on a ext2 filesystem can be between 1024 and 4096.
>
>would you care to explain what block size you are referring to and
>what block sizes other systems where Oracle is available use?
>
>| 6. Linux SMP is rudimentary and flaky at best.
>
>that might be true, but still, they ported Oracle to NT and last time
>I checked Linux scaled better across 2 CPUs than NT.
>
>would you care to explain?
>
>| 7. The same for multithreading.
>
>from what I understand Linux multithreading is different from for
>example Solaris. I'm told (by people who are intimately familiar
>with the Linux kernel) that Linux does multithreading faster than for
>example Solaris.
>
>it's there, it's fast -- now, would you care to explain exactly why
>you think that Linux has "rudimentary and flaky at best"
>multithreading?
>
>| In fact porting Oracle to Linux would require Oracle to write its
>| own Linux almost from scratch.
>
>you are most absolutely wrong.
>
>| ==> I don't see much (or rather _any) prospects for seeing Oracle on
>| Linux.
>
>perhaps it's time for you to take a closer look at Linux as well as
>other systems that Oracle has been ported to before you jump to any
>conclusions about the suitability of Linux as a platform for running
>Oracle. Oracle has been ported to operating systems that are _far_
>less advanced than Linux and that lack almost everything that you
>mentioned in your post. that hasn't stopped Oracle from porting to
>them so far, so you will have to think of something better.
Jeez, I had to support it on MPE/ix! Fortunately, not for any actual
customers.
>
>the fact that Oracle already can be run on Linux using iBCS2 (this has
>been possible since at least version 1.2 of the Linux kernel) should
>prove that Oracle can in fact run under Linux without, as you state,
>Oracle Corp. having to rewrite Linux from scratch.
>
>(Some time ago I read that someone had benchmarked Oracle on a PC using
> both SCO UNIX and Linux with some very interesting results. does
> anyone remember who did this?)
>
>
>porting Oracle to Linux is certainly not a technical issue, it is a
>political and economical one. if the CEO of Oracle is as dedicated to
>the cause of fighting Microsofts operating system monopoly as he would
>like us to think he is it puts him in a strange light when he doesn't
>pay attention to those who use one of the more viable alternatives
>to NT turn to Oracle for support.
Yeah, that business plan youse guys were talking about - try packaging
linux servers with a few million of NC's. That oughta get Larry's attention.
>
>
>-Bjørn
Evan Leibovitch
In article <67cqc4$c...@bgtnsc03.worldnet.att.net>,
Nilanjan Bhowmik <ni...@worldnet.att.net> wrote:
>Josue Emmanuel Amaro wrote:
>> Lets face it, management at Oracle will simple answer with "show me the money".
>> Believe me, if someone can come up with a business plan with some hard
>> numbers, maybe someone will listen.
>If we can create a solid business plan and may be small presentation
>( anyone here has pull in ORACLE ? ) to them I think they will consider
>it. We may need to involve Caldera in it - they may be trying it by now
>but it does not harm to give them support for it.
I can't speak specifically for Oracle, but I had a long conversation with
someone at Progress about their DBMS's lack of Linux support. Amongst the
comments to arise:
- Technical porting issues are not the obstable, support is the biggie
We would have to be able to guarantee *huge* revenue to change their
minds on this
- Most DBMS makers are trying to *reduce* the numbers of different
ports/versions, they getting choked by the large numbers of SKUs
they now muct deal with and support
- Progress has no objection to Linux per-se, they just object to having
to make and support yet one more native binary product, and they
won't support anything that runs under emulation. If SCO and Linux
could run the same native binary Progress would (they told me) have
no objection to supporting both platforms with that binary.
These comments are amongst the things that got me involved in the 86open
project (http://www.telly.org/86open). The fact is that SCO, Linux,
SolarisX86 and the BSDs all have binary formats that are oh-so-similar
but just different enough to be a pain, and this diversity presents a
significant obstacle to anyone looking to port to Unix-on-Intel.
One more point specifically on Josue's comment:
The HUGE difficulty in presenting a business plan regarding the
acceptance of Linux, is that there is *still* no really accurate
way to indicate to software vendors just how many Linux users are
out there, let alone how many would be willing to buy an expensive
DBMS.
The IDGs and other research firms on which the industry and its
media depend, won't admit Linux exists until someone pays they to
do it. Linux doesn't even show up on current breakdowns of the
Unix market -- how is an ISV to know just how big this market is?
Furthermore, just how much of this market is prepared to spend
significant cash on applications, considering that it hasn't spent
much on the OS and significant utlities.
Yes, I know Caldera sells a version of Linux for $399, and businesses
who buy that clearly indicate a willingness to pay for software -- but
how much of the whole Linux world is that?
These are questions that must be asked and answered before many ISVs
will get involved -- not just Oracle.
--
Evan Leibovitch, Sound Software Ltd, located in beautiful Brampton, Ontario
Supporting PC-based Unix since 1985 / Caldera & SCO authorized / www.telly.org
When the only tool you have is a hammer, every problem comes to look like a nail
Bjorn Borud
[joe...@pebble.ml.org (Joel Garry)]
|
| With the top-end MP PC's, PC is a misnomer. I don't think parallel
| servers are a bad topic for this at all. Too bad the original post
| was such a troll.
the problem with PCs is that you still have to be very careful when
building them because compared to traditional UNIX-server hardware
they are still not very robust. I've had disks litterally burn up
inside the cabinet because cooling is usually bad and my dual P6 needs
about 3 new cooling fans per year.
the problems with PCs is usually shitty hardware.
but as far as performance is concerned there is nothing near the price
ranges of PCs that gives you more bang for the buck. often I
preprocess data on my workstation instead of the database machine,
because the Intel running Linux is several times faster.
| Yeah, that business plan youse guys were talking about - try
| packaging linux servers with a few million of NC's. That oughta get
| Larry's attention.
do we know anyone who knows Larry personally and could give him a call
and explain to him what Linux is and that there are some people who
would really like to run Oracle on it? :-)
PS: I'd like to run Designer/2000 on Linux as well :-)
tip
there is NO WAY the pc hardware end of linux would be able to handle running
oracle in any decent manner. plus its tweaking is rather limited compared to
solaris and hpux. i have run oracle on both suns and hp boxes - there's just
no way linux is ready (yet) for oracle.
i currently run oracle 7.2.3 on hp9000/s800/k210's running hpux 10.10 with a
gig of memory each. that's alright for running oracle.
but imagine a pc with linux and a motherboard limitation of 128M, a 200Mhz
processor... i don't think so...
linux is good for being a workstation, or a small server - but the heavy duty
shit - better leave it to superior hardware and os's.
thanks,
- tip
In article <m2yb1ep...@lucifer.guardian.no>,
Bjorn Borud <bo...@guardian.no> wrote:
>[joe...@pebble.ml.org (Joel Garry)]
>|
>| With the top-end MP PC's, PC is a misnomer. I don't think parallel
>| servers are a bad topic for this at all. Too bad the original post
>| was such a troll.
>
>the problem with PCs is that you still have to be very careful when
>building them because compared to traditional UNIX-server hardware
>they are still not very robust. I've had disks litterally burn up
>inside the cabinet because cooling is usually bad and my dual P6 needs
>about 3 new cooling fans per year.
>
>the problems with PCs is usually shitty hardware.
>
>but as far as performance is concerned there is nothing near the price
>ranges of PCs that gives you more bang for the buck. often I
>preprocess data on my workstation instead of the database machine,
>because the Intel running Linux is several times faster.
>
>| Yeah, that business plan youse guys were talking about - try
>| packaging linux servers with a few million of NC's. That oughta get
>| Larry's attention.
>
>do we know anyone who knows Larry personally and could give him a call
>and explain to him what Linux is and that there are some people who
>would really like to run Oracle on it? :-)
>
>
>PS: I'd like to run Designer/2000 on Linux as well :-)
>
>-Bjørn
>--
> Bjørn Borud <bo...@guardian.no> | "The Net interprets censorship
> <URL:http://www.pvv.unit.no/~borud/> | as damage and routes around it."
> UNIX person, one of "them" | - John Gilmore
--
[ =-=-= want to provide financial backing for my film? email me! -=-= ]
[ *&$*&$* tip - my evil twin is pit - mailto:t...@blahblah.com *&$*&$* ]
[ & email is spam protected: replace "blahblah" with "stopsmiling" &* ]
[ =-=-=-=-=-= preserve wildlife; pickle a squirrel today. =-=-=-=-=-= ]
Stephen Harris
Bjorn Borud (bo...@guardian.no) wrote:
: ["S V" <s...@mindspring.com>]
: | 1. Linux has no logical volumes layer.
: what's the problem? I have worked on ONE site where they used logical
: volumes, but they would have done just as well without. this is a non
: issue.
Even more so - all competent DBA's I know of *hate* logical volumes. DBA's
like to know exactly what disk spindle and what SCSI controller their data
file is on, so they can make an reasoned guess at distributing IO load
evenly over controllers/spindles. LVM structures (eg raid0 or raid5) destroy
that.
However, the logical volume structure of HPUX and Solaris ODS isn't much more
(if any more!) than the metadisk driver currently in Linux.
: | 3. Linux has no support for raw devices - hence NO even remote possibility
: | to run Oracle Parallel Server.
: I have yet to desire raw devices for using Oracle and I certainly have
: never needed to run Oracle Parallel server. if I would ever need to
: run it I would certainly not do it on a PC.
Emphasised! If you need OPS, then you are not going to go anywhere near a
PC server. You're talking HPUX Kclass servers with MC/Serviceguard or Solaris
E3000's with SSAs at a minimum.
: | 4. Linux networking is flaky at best.
: uh, compared to what? would you mind explaining to me how you got
Linux networking is very good actually. Perhaps SV is thinking of the 0.99pl12
kernel :-)
: | 5. Linux OS block size is what? 512 bytes? It would make even MS
: | Access laugh and puke steam.
: uh, what block size would that be? if you're talking about the file
Err, this is also irrelevant since database systems like Oracle have
tuning features to allow block pre-fetch, multi-block fetchs etc in one IO
operation.
: | 6. Linux SMP is rudimentary and flaky at best.
: that might be true, but still, they ported Oracle to NT and last time
: I checked Linux scaled better across 2 CPUs than NT.
Linux 2.0 SMP isn't too hot (lots of mutex locking) but works. 2.1 SMP is
a lot better and faster.
: | 7. The same for multithreading.
: from what I understand Linux multithreading is different from for
: example Solaris. I'm told (by people who are intimately familiar
Err, Oracle didn't even properly do Multi-threading until 7.3.3, and doesn't
really take full advantage of it until version 8.0.3. Indeed, for a lot of
applications, the MTS options _SLOW_DOWN_ the database on HPUX and Solaris
systems. Your requirements have to be quite specialised to benefit from
enabling MTS in Oracle. [ This is what my DBA told me after coming back
from the Oracle Performance Tuning course, upon which he promptly turned off
the MTS options! ]
: | In fact porting Oracle to Linux would require Oracle to write its
: | own Linux almost from scratch.
: you are most absolutely wrong.
Yup!
: | ==> I don't see much (or rather _any) prospects for seeing Oracle on
: | Linux.
: perhaps it's time for you to take a closer look at Linux as well as
: other systems that Oracle has been ported to before you jump to any
Well, SV is right, but for the wrong reasons. Oracle rarely jumps at things
unless they see $$$$$$$
--
Stephen Harris
sw...@spuddy.mew.co.uk http://www.spuddy.org/
The truth is the truth, and opinion just opinion. But what is what?
* Meeeeow ! Call Spud the Cat on > 01268 515441 < for free Usenet access *
Stephen Harris
Bjorn Borud (bo...@guardian.no) wrote:
: | Oracle needs character devices for disk partitions.
: you don't really need raw devices; it's a feature you can live without
: in most cases. if you need them you simply run Oracle on some other
: platform. usually the OS will have the best idea about how the I/O
Ugh, every follow up I've seen here has missed the point. RAW device access
is unbuffered. This is *essential* if you want the same disk to be accessable
by two machines at the same time, which is essentially what running Oracle
Parallel Server entails.
However, it should be noted:
1) OPS doesn't work very well. At work we have spoken to Oracle re
implementing an OPS solution integrated with HP's MC/Serviceguard. After
a few months Oracle came back and said "we give up". Oracle weren't even
able to show us any UK reference sites for us to look at. We gave up and
went to a log-shipping solution :-)
2) OPS is a heavy duty solution, and you wouldn't even go near it with a
Linux box (or a SCO box, or an NT box) if you were halfway sane. We're
talking the province of big servers here :-)
So, raw disk access isn't that much of a problem :-)
Jim Zubb
tip wrote:
> there is NO WAY the pc hardware end of linux would be able to handle running
> oracle in any decent manner. plus its tweaking is rather limited compared to
> solaris and hpux. i have run oracle on both suns and hp boxes - there's just
> no way linux is ready (yet) for oracle.
>
This is just a stupid comment, Oracle is available for Netware and NT, bothx86
based OS's, just like Linux, plus Linux can run on Alpha, PPC, Mips, etc...
So how exactly is the PC hardware of the Netware and NT boxes so much
better that they can run Oracle?? They are the same hardware as Linux!
Granted most of the hardware running Linux is lower end, but it is more than
capable of running on the high end stuff...it is just tough to convince the boss
to
let a free OS run his $10k server...
--
Jim Zubb
ji...@ecom.net
Uwe Bonnes
In comp.os.linux.development.apps tip <t...@blahblah.com> wrote:
: there is NO WAY the pc hardware end of linux would be able to handle running
: oracle in any decent manner. plus its tweaking is rather limited compared to
: solaris and hpux. i have run oracle on both suns and hp boxes - there's just
: no way linux is ready (yet) for oracle.
: i currently run oracle 7.2.3 on hp9000/s800/k210's running hpux 10.10 with a
: gig of memory each. that's alright for running oracle.
: but imagine a pc with linux and a motherboard limitation of 128M, a 200Mhz
: processor... i don't think so...
: linux is good for being a workstation, or a small server - but the heavy
: duty shit - better leave it to superior hardware and os's.
Linux runs on ALPHA,MacLinux, Mips, ...
Don't think on I86 only.
Bye
--
Uwe Bonnes b...@elektron.ikp.physik.tu-darmstadt.de
Free Software: Contribute nothing, expect nothing
--
John Hagelgans
tip wrote:
> there is NO WAY the pc hardware end of linux would be able to handle running
> oracle in any decent manner. plus its tweaking is rather limited compared to
> solaris and hpux. i have run oracle on both suns and hp boxes - there's just
> no way linux is ready (yet) for oracle.
This is really quit funny since here I am running a 300 user Sybase server on
a dual 100 MHz Pentium running NCR Unix. Granted the reason is historical since we
are were a division of AT&T once, but the point is that the Pentium can
handle the load, most of the time.
The real reason that it works is that we are using SCSI disks 256 Meg of memory (maxed or I'd have more). Thankfully, we are moving to HP's in the near future. Still, I am convinced that t
he problem with PC's is not a hardware issue, everyone has become so used to the shitty
response of the hardware with NT and 95 that you kind of feel liberated when you
boot up a real operating system instead of a TOY! The real reason everyone thinks the
hardware sucks is that the software is so bad. Intel will be wise to dump MS on the
next go-round and that may just happen.
Linux is equivalent at least to NCR Unix and thus I'd have to say running Oracle on
Linux would not be too much of stretch. If not maybe someone can convince Sybase to
port to Linux.
-JH
Kevin Huber
tip <t...@blahblah.com> writes:
> there is NO WAY the pc hardware end of linux would be able to handle running
> oracle in any decent manner. plus its tweaking is rather limited compared to
> solaris and hpux. i have run oracle on both suns and hp boxes - there's just
> no way linux is ready (yet) for oracle.
In what sense can't you tweak Linux? You have the full source code...
-Kevin
Roope Anttinen
tip <t...@blahblah.com> wrote:
> there is NO WAY the pc hardware end of linux would be able to handle running
> oracle in any decent manner. plus its tweaking is rather limited compared to
> solaris and hpux. i have run oracle on both suns and hp boxes - there's just
> no way linux is ready (yet) for oracle.
How come Oracle is running even on some Win 3.x PC's???
> i currently run oracle 7.2.3 on hp9000/s800/k210's running hpux 10.10 with a
> gig of memory each. that's alright for running oracle.
Depends on the level of running..
> but imagine a pc with linux and a motherboard limitation of 128M, a 200Mhz
> processor... i don't think so...
What!!
cdlinux01 <22.12 11:15 ~> free
total used free shared buffers
Mem: 256812 230780 26032 77788 87880
-/+ buffers: 142900 113912
Swap: 0 0 0
MemTotal: 250
MemFree: 25
MemShared: 75
Buffers: 85
Cached: 60
SwapTotal: 0
SwapFree: 0
cdlinux01 <22.12 11:15 ~> uname -a
Linux cdlinux01 2.1.72 #5 Thu Dec 11 20:22:45 EET 1997 i686 unknown
Tell this PC motherboard that it can't take over 128M of RAM! and have you
heard of Pentium II's?
What about alpha? It could be conscidered to be PC too. Is that too flakey
to run Oracle with Linux? Have you heard of Linux been ported to PA-RISC,
Sparc etc...
Roope
--
MicroSoft? is that some kind of a toilet paper?
PS: Look for address here, not from headers. And remove NOSPAM's
___________________________________________________________________________
antt...@myy.NOSPAM.helia.fi / Roope.A...@ntc.NOSPAM.nokia.com
+358 9 812 7567 / +358 49 465 902 / +358 49 445 565
http://myy.helia.fi/~anttiner/index.html
---------------------------------------------------------------------------
Helsinki Business Polytechnic - Institute of information technology
John S. Dyson
In article <m2yb1ep...@lucifer.guardian.no>,
Bjorn Borud <bo...@guardian.no> writes:
>
> do we know anyone who knows Larry personally and could give him a call
> and explain to him what Linux is and that there are some people who
> would really like to run Oracle on it? :-)
>
profits that could be made. :-). Take a look at the address in my
signature line.
--
John | Never try to teach a pig to sing,
dy...@freebsd.org | it just makes you look stupid,
jdy...@nc.com | and it irritates the pig.
Hannu Krosing
tip wrote:
> there is NO WAY the pc hardware end of linux would be able to handle running
> oracle in any decent manner. plus its tweaking is rather limited compared to
> solaris and hpux.
Sorry for may ignorance, but what is tweaking ?
And what is wrong with linux implementation of it?
> i have run oracle on both suns and hp boxes - there's just
> no way linux is ready (yet) for oracle.
What's so wrong with oracle that it's not ready for linux? Linux is widely
recognized as one of easiest platforms to port to.
> i currently run oracle 7.2.3 on hp9000/s800/k210's running hpux 10.10 with a
> gig of memory each. that's alright for running oracle.
Actually you can run oracle on much smaller machines.
> but imagine a pc with linux and a motherboard limitation of 128M, a 200Mhz
> processor... i don't think so...
You could also get a PC with a motherboard limitation of 2G and 2 300MHz
processors.
Or even more if you get Alphas
> linux is good for being a workstation, or a small server - but the heavy duty
> shit - better leave it to superior hardware and os's.
Oracle is quite usable for not-so-heavy-duty shit as well. It is used even on NTs
and Netware servers, not to mention Win95 boxes.
Hannu
Tor Arntsen
In article <m267ol3...@lucifer.guardian.no>,
Bjorn Borud <bo...@guardian.no> writes:
>["S V" <s...@mindspring.com>]
>|
>| And in any case we discuss the native Oracle port, not the one
>| which runs under SCO libraries.
>
>the iBCS2 interface merely remaps system calls and provides wrappers
>to some system calls in order to provide the semantics of the OS it is
>supposed to mimic.
>
>the point is: Oracle can run under Linux so there is no technical
>reason it shouldn't be ported. also I don't think that it would be
>that much of a job either given that Oracle is written to be portable
>and it has been ported to stranger OSes than Linux.
According to people that have been in contact with Oracle they *do*
run Oracle natively on Linux, internally in their own porting department.
The reason they don't sell a native Linux version is then obviously not
for technical reasons, but purely for marketing reasons. Centralised
commercial Linux support and all that. We all know the real truth about
that of course, in any case it's obviously up to Oracle what they want
to do with their product.
Those who want it could try to pool up and come up with a *big* order,
I've seen people trying to organise something like that.
And that guy "S V" is either a troller or just stupid.
Tor
Craig Kelley
In article <67jrfn$h...@eve.enteract.com>, tip <t...@blahblah.com> wrote:
->there is NO WAY the pc hardware end of linux would be able to handle running
->oracle in any decent manner. plus its tweaking is rather limited compared to
->solaris and hpux. i have run oracle on both suns and hp boxes - there's just
->no way linux is ready (yet) for oracle.
Bullshit. We have an AIX 3.2.5 machine (circca 1994) running Oracle
7.3 and our dual-pentium Pro machine with 256MB RAM runs CIRCLES
around the machine.
->i currently run oracle 7.2.3 on hp9000/s800/k210's running hpux 10.10 with a
->gig of memory each. that's alright for running oracle.
->
->but imagine a pc with linux and a motherboard limitation of 128M, a 200Mhz
->processor... i don't think so...
Ummm, get a quad motherboard which supports a GB of RAM-- it'll beat
the 9000 in raw CPU performance (it will not have as good I/O scores);
We have both. The only thing holding the x86 machine down is the lack
of a good OS + DBMS. (which could be Linux + Oracle or Sybase)
->linux is good for being a workstation, or a small server - but the heavy duty
->shit - better leave it to superior hardware and os's.
Whatever. I can easily get four machine of the same/better
performance for the price of an HP9000.
I wonder who is going to win.. If Oracle or Sybase puts their DMBS on
Linux they will get a nice short-term cash flow as initial purchases
are made-- the other isn't. :)
--
ASCII stupid question, get a stupid ANSI
Craig Kelley -- kell...@isu.edu
http://www.isu.edu/~kellcrai finger i...@inconnu.isu.edu for PGP block
tip
In article <67lbbh$b...@myy.helia.fi>, Roope Anttinen <nos...@foo.bar> wrote:
>
>How come Oracle is running even on some Win 3.x PC's???
are you saying their performance is great?
any os can run a particular kind of app, but the question is does it run well?
>Depends on the level of running..
exactly.
>> but imagine a pc with linux and a motherboard limitation of 128M, a 200Mhz
>> processor... i don't think so...
>
>What!!
>cdlinux01 <22.12 11:15 ~> free
> total used free shared buffers
>Mem: 256812 230780 26032 77788 87880
MOST motherboards have a 128M limit.
i didn't say all.
tip
In article <349E0D01...@spam.sux>, Jim Zubb <j...@spam.sux> wrote:
>
>Granted most of the hardware running Linux is lower end, but it is more than
>capable of running on the high end stuff...it is just tough to convince the boss
>to
>let a free OS run his $10k server...
same thing as saying a 486 with 16megs can run photoshop.
Thomas Schenk
tip wrote:
>
> there is NO WAY the pc hardware end of linux would be able to handle running
>
>
> processor... i don't think so...
>
>
This is total crap. The entire Deja News service is run on Linux
machines
and Deja News has one of the largest databases on the Web. I also don't
know what decade you are living in, but we have Linux SMP running on
dual
Pentium II, 300Mhz, with 512M and 28 Gigs of hard disk space, so that
stuff
about 128M limitation and 200Mhz is bogus. I am not saying that Linux
is
the end all and be all of operating systems, but don't write it off as
inferior simply because it runs on PCs. Just because it isn't a PC
doesn't
make it superior hardware.
Tom
--
+-----------------------+-------------------------+------------------------+
| Tom Schenk | Use Linux! | All opinions
expressed |
| tsc...@dejanews.com | Friends don't let | are mine and not
those |
| tsc...@theoffice.net | friends do Windows! | of my
employer. |
+-----------------------+-------------------------+------------------------+
Joel Garry
In article <67jrfn$h...@eve.enteract.com>, tip <t...@blahblah.com> wrote:
>there is NO WAY the pc hardware end of linux would be able to handle running
>oracle in any decent manner. plus its tweaking is rather limited compared to
>solaris and hpux. i have run oracle on both suns and hp boxes - there's just
>no way linux is ready (yet) for oracle.
>
>i currently run oracle 7.2.3 on hp9000/s800/k210's running hpux 10.10 with a
>gig of memory each. that's alright for running oracle.
>
>but imagine a pc with linux and a motherboard limitation of 128M, a 200Mhz
>processor... i don't think so...
I'd rather not think about that ... :)
>
>linux is good for being a workstation, or a small server - but the heavy duty
>shit - better leave it to superior hardware and os's.
I used to be really anti-intel, but some of these newer Pentiums finally have
convinced me, as you say, good for a small server.
>
>thanks,
>- tip
>
>In article <m2yb1ep...@lucifer.guardian.no>,
>Bjorn Borud <bo...@guardian.no> wrote:
>>[joe...@pebble.ml.org (Joel Garry)]
>>|
>>| With the top-end MP PC's, PC is a misnomer. I don't think parallel
>>| servers are a bad topic for this at all. Too bad the original post
>>| was such a troll.
>>
>>the problem with PCs is that you still have to be very careful when
>>building them because compared to traditional UNIX-server hardware
>>they are still not very robust. I've had disks litterally burn up
>>inside the cabinet because cooling is usually bad and my dual P6 needs
>>about 3 new cooling fans per year.
>>
>>the problems with PCs is usually shitty hardware.
I certainly agree with that! However, just because most PC's are poorly
engineered, or perhaps engineered to a price point lower than us
technoweenies appreciate, doesn't mean a top-end MP Intel server has to
be bad. I've seen 64 bit HP and Alphas screw up, too, it's just not
something most people have to deal with - and they send engineers out to
fix 'em, because they are usually under an expensive support contract
or development agreement.
While I personally would rather have a "big" sun, hp or alpha, I haven't seen
anything very convincing that a decent mp Intel running linux and parallel
Oracle would be a bad idea for, say, a deparmental/small-office server, were
such a beast to work.
>>
>>but as far as performance is concerned there is nothing near the price
>>ranges of PCs that gives you more bang for the buck. often I
>>preprocess data on my workstation instead of the database machine,
>>because the Intel running Linux is several times faster.
>>
>>| Yeah, that business plan youse guys were talking about - try
>>| packaging linux servers with a few million of NC's. That oughta get
>>| Larry's attention.
>>
>>do we know anyone who knows Larry personally and could give him a call
>>and explain to him what Linux is and that there are some people who
>>would really like to run Oracle on it? :-)
Well, one of the bay area papers had an article about his landscape
architect... :)
>>
>>
>>PS: I'd like to run Designer/2000 on Linux as well :-)
>>
>>-Bjørn
>>--
>> Bjørn Borud <bo...@guardian.no> | "The Net interprets censorship
>> <URL:http://www.pvv.unit.no/~borud/> | as damage and routes around it."
>> UNIX person, one of "them" | - John Gilmore
>
>
>--
>[ =-=-= want to provide financial backing for my film? email me! -=-= ]
>[ *&$*&$* tip - my evil twin is pit - mailto:t...@blahblah.com *&$*&$* ]
>[ & email is spam protected: replace "blahblah" with "stopsmiling" &* ]
>[ =-=-=-=-=-= preserve wildlife; pickle a squirrel today. =-=-=-=-=-= ]
Joel Garry
In article <67maho$b...@eve.enteract.com>, tip <t...@blahblah.com> wrote:
>In article <349E0D01...@spam.sux>, Jim Zubb <j...@spam.sux> wrote:
>>
>>Granted most of the hardware running Linux is lower end, but it is more than
>>capable of running on the high end stuff...it is just tough to convince the boss
>>to
>>let a free OS run his $10k server...
>
>same thing as saying a 486 with 16megs can run photoshop.
???
I do that?
The hardest part was finding damn drivers for the video card/monitor.
jg
Benjamin Redelings I
Err... I notice lots of people saying things like "PC hardware sucks" or
talking about 128Mb memory limitations. This is ridiculous...
PPro:
Largest Cacheable Area: 4Gb
SMP: Can handle up to 4 Processors.
On-chip cache: 256k,512k,1024k
Mhz: 150,166,180,200
As far as motherboards go, I'd be careful before saying "there aren't
any PPro that do X." For example, you might think that "there aren't
any PPro motherboards with 75Mhz bus" or "there aren't any PPro
motherboards with SD-RAM." You'd be wrong! They just aren't very
common :)
-BenRI
Jay Thorne
Thomas Schenk wrote:
>
> tip wrote:
> >
> > there is NO WAY the pc hardware end of linux would be able to handle running
> > oracle in any decent manner. plus its tweaking is rather limited compared to
> > solaris and hpux. i have run oracle on both suns and hp boxes - there's just
> > no way linux is ready (yet) for oracle.
> >
> > i currently run oracle 7.2.3 on hp9000/s800/k210's running hpux 10.10 with a
> > gig of memory each. that's alright for running oracle.
> >
> > but imagine a pc with linux and a motherboard limitation of 128M, a 200Mhz
> > processor... i don't think so...
> >
> > shit - better leave it to superior hardware and os's.
> >
>
> machines
> and Deja News has one of the largest databases on the Web. I also don't
> know what decade you are living in, but we have Linux SMP running on
> dual
> Pentium II, 300Mhz, with 512M and 28 Gigs of hard disk space, so that
> stuff
> about 128M limitation and 200Mhz is bogus. I am not saying that Linux
> is
> the end all and be all of operating systems, but don't write it off as
> inferior simply because it runs on PCs. Just because it isn't a PC
> doesn't
> make it superior hardware.
>
that much machine since oracle is so horribly slow. Its slow because of
all the ridiculous things that that come with it. Transactions,
rollbacks, logs, PL/SQL. Eeeew.
I've benchmarked other databases against Oracle, and its laughably
slow. So why do you need a Gig of ram on a huge HP box? With a pair of
PII's@300Mhz, 512MRam, Oracle will be reasonable speed, but a memory
pig. Except for the largest transaction oriented banking systems, 512M
is pretty useful. PII's at 300Mhz are faster than the big HP, benchmark
wise. PCI is a bit slow for really huge memory & IO, but its within the
right order of magnitude. The big SPARKs Alphas and HP boxes have twice
or more the memory/IO bandwidth. You need that when your app starts to
use a ton of RAM & Disk.
With Solaris on a Spark 20 with 256M of ram, one medium sized oracle web
app grows to 120 M ram in 3 days. Puke. This is why the guy needs a Gig
of RAM. I cross ported to another DB(Texpress) on Linux/x86 and the app
is 5M after 30 days.
From my experience, Linux would get an extra 10 to 20% out of the
hardware compared to any of the alternatives. Its easier to port to
than SCO, tons faster than Solaris and SCO and NT, and way more reliable
than NT or SCO. Solarix X86 we had no reliability issues with , nor
with Linux, but we did with NT and SCO. My main linux server has been
down 6 times in the last year, each of them for OS upgrades.
In a bank/high volume commercial/governmental situation, you need
extremely reliable Hardware and Software. Thats what all the
"Ridiculous" extra transaction features of Oracle are for. If I was
suddenly a CIO at a bank, I would not consider running the credit card
system on Linux. Not because of OS reliability, but because the
stockholders would lynch me if my datacenter was not supported by a
reliable company. The OS doesnt even enter into it. The hardware is
the basic unit of reliability here. Oracle enters into it, only because
you need serious hardware to get good speed because its so slow.
I'd be a dyed in the wool big iron guy (IBM,DEC,Sun,HP). Not only is the
_MACHINE_ a different breed, (at $5Million for a 3090/600e or $400K for
a really huge DEC Alpha with enough CPU/Disk/RAM/Support contract) but
the _SUPPORT_ the vendor gives the bank is of a different breed. When
you are an IBM, a Digital, or an HP or a Sun, you put a hardware/OS
support guy on-site. No Linux vendors can do that, becase the
profitability per shipped unit is not enough.
A $400,000 Dec box is quite a bit different of a machine than even a
HUGE PC architecture server. Even compared to a $30K tricked out
Intergraph or Compaq box. Big iron is more reliable, hardware wise.
Its much more reliable software wise, because for $400K, you get the
time to check the MTBF's properly, and regression test the OS on the
actual hardware you ship in all sorts of horrible conditions. You test
it with poor power and lots of electrical noise and radio noise around.
Until the NEED for a really huge centralized PC architecture box
happens, there will not be a PC machine of the big iron reliability
class either. There will be no need, because the prevailing wisdom is
to put in another server, rather than buy a bigger single server. Its
tremendously cheaper for most offices.
Now, if you need to have a huge central system (say you are Chase
Manhattan) then you talk to vendors who make money from support, not
from hardware/OS sales.
Said the Ex DEC Field Service engineer, who spent most of the Eighties
in cold computer rooms with his hands in the guts of VAXes. (Where
Field service was a profit center).
+-----------------------+-------------------------+------------------------+
> | Tom Schenk | Use Linux! | All opinions
> expressed |
> | tsc...@dejanews.com | Friends don't let | are mine and not
> those |
> | tsc...@theoffice.net | friends do Windows! | of my
> employer. |
> +-----------------------+-------------------------+------------------------+
--
Jay Thorne The Net Result System Services j...@result.com
Http://net.result.com
Zoom 505 Effect page http://net.result.com/~jay
Zoom 5xx series Patch Database: http://net.result.com/~jay/db.html
tip
In article <349fcd70....@news.mindspring.com>,
Voytek Jarnot <vja...@mindspring.com> wrote:
>
> You are so full of it that it hurts me to waste the time to reply.
>What exactly does 'most' mean to you??? Here's a breakdown of some of
>intel's various chipsets, its the chipset that the motherboard
>incorporates, not the motherboard itself, that limits the amount of
>RAM.
are you blind or just dumb? what does linux run on? 386, 486, pentium,
whatever.
i'm not talking about the later model pci chipset motherboards in the
past two years, rather as a whole.
as a whole, most motherboards that support linux do not support over
128 megs.
Benjamin Redelings I
Wow, thanks for all the info. That was rather interesting :)
Still, as to the question of "Why no Oracle on Linux?", I think its
probably been answered pretty well, and I don't think its really a Linux
issue. Given some of the OTHER machines its been ported to, I suspect
that complaining about Linux isn't really the issue (tho taking into
account the things you said about hardware).
However, why Oracle? I'm not really a transaction-oriented-database
person (Deductive Databases are cool :), but WHY Oracle? Someone said
that it's marketing was better than its product... How about other ones?
I do have ONE problem here though:
> In a bank/high volume commercial/governmental situation, you need
> extremely reliable Hardware and Software. Thats what all the
I suspect that not everyone who uses Oracle is in this postion.
However, perhaps if they are not, then they shouldn't use Oracle :)
Thanks,
-BenRI
Michael Hasenstein
Right.
There was a sub-thread about traditional Unix-workstations vs Intel-WS,
mostly saying that the Intel stuff (well, the components and the system
architecture) is of (much?) lower quality. While this may be true for
most or even all machines bought for low $ at the nearest computer shop
I'd say it isn't for the big Intel servers sold by e.g. Compaq or
Siemens or by (whoever) for running Windows NT (up to 8 processors by
now, how NT scales (or not) is a different story). How many databases
are there that need a 24 CPU server with 24 GB RAM and some TeraByte
disk space? Of course, companies running these are our "most valuable"
customers but guess why Oracle is available on NT, and I'd guess Linux
has at least the same potential, porting should even be easier (to Linux
vs to NT; never tried so you can flame me for that statement - there are
so many similar bad ones in the news from people who write about things
they never tried).
A Primergy 760-server with 4 PPro's and 4GB RAM
(http://www.sni.de/servers/primergy/prim_us/pr760_us.htm, hope this
doesn't make my article commercial spam, that's the one I happen to
know; just as an example for those who only know home-PC's that there's
more/bigger Intel stuff out there) should be enough for most uses.
A note: the From:-address doesn't mean I know much or anything about
Oracle's strategy here,
--> I'm just a guest here @Oracle HQ <--
(for 6 months) doing a 64bit port of Oracle 8 for our Reliant-Unix
(SysV, Siemens Nixdorf). Who would come here for doing this on Linux?
All the vendors who want to have Oracle available on their platform have
a couple of people here to help with porting and to provide a link back
to the people who code the OS. There are so many issues that can best be
solved by the platform's vendor, e.g. there are changes made to the OS
(rather than Oracle) in order to support certain features better (e.g.
make some things faster for the particular Oracle-implementation -
easier than rewriting Oracle, sometimes, also see below).
Oracle is not so high level like, let's say, a text editor, it relies
and uses low level functionality a lot. When doing a port, sometimes you
adjust the OS's behaviour or functionality in order to support an Oracle
function better, not Oracle itself (often because they did it right or
because all others already have that OS-functionality, so one can't
afford not to do it). For this it's good to have an organization to talk
to, so I'd say if Caldera or someone else wants to be the partner for
Oracle, they probably have a much better chance than the anonymous and
fuzzy crowd of Linux users out there (include me there). Either Oracle
can be convinced that it pays to do the port or somebody else takes the
risk and does it, here at Oracle and together with Oracle. I don't see
why _Oracle_ should take the risk. I'm a Linux user for years (even did
some kernel programming, see URL in sig) and have used it in the
company (was: debis (Daimler Benz), now: Siemens) as often as I could
(which was a lot), but I'd not take that risk because I have absolutely
no idea if I could actually _sell_ the result of my efforts often enough
(and I really can't see anything bad in that Oracle wants to make money
- how long would they (or any other company) exist if they stopped
looking for profit?).
Other OS's have an easier life here: there's just one vendor who
produces it and all wishes for a certain product go there, so it can
easily be seen wether or not spending money on a port or development can
be justified. For Linux there are only a few postings (and not even that
many) in newsgroups, and I doubt that a large percentage of people who
said there they'd like to have a version on Linux would really spend
lots of money (and that's what commercial DBs cost, especially if you
have many users: lots of money! (and that's also what must be spent for
the development of such a product)) on it when it's available, because
when they see the sum they might think buying a big Unix server (or NT
nowadays, unfortunately [I should add in these groups - also
unfortunately, although I don't like NT very much there's too much heat
in these discussions and less substance (I said less, not 'no'! Stop
flaming!)]) doesn't matter anyhow.
I realize I wrote lots of unreadable sentences (too long, too many
'(')... Please excuse me, it's late and I want to go home.
This article is long enough now, so I just press 'Send' before I can
write even more...
--
***********************************************************************
Michael Hasenstein; Siemens Nixdorf (Paderborn, Germany)
currently @Oracle, Redwood Shores, California
homepage http://www.csn.tu-chemnitz.de/home/mha/
Linux NAT http://www.csn.tu-chemnitz.de/HyperNews/get/linux-ip-nat.html
private EMail: M.Hasenstein(at)remove-this rocketmail dot com
***********************************************************************
Tim Smith
Benjamin Redelings I <bred...@ucsd.edu> wrote:
> However, why Oracle? I'm not really a transaction-oriented-database
Oracle (the company) does not like Microsoft. Therefore, at least the way
some Linux users think, Oracle's products must be cool and they want them
on Linux.
--Tim Smith
Jim Zubb
tip wrote:
> In article <349fcd70....@news.mindspring.com>,
> Voytek Jarnot <vja...@mindspring.com> wrote:
> >
> > You are so full of it that it hurts me to waste the time to reply.
> >What exactly does 'most' mean to you??? Here's a breakdown of some of
> >intel's various chipsets, its the chipset that the motherboard
> >incorporates, not the motherboard itself, that limits the amount of
> >RAM.
>
> are you blind or just dumb? what does linux run on? 386, 486, pentium,
> whatever.
>
> i'm not talking about the later model pci chipset motherboards in the
> past two years, rather as a whole.
>
> as a whole, most motherboards that support linux do not support over
> 128 megs.
>
This has got to be the dumbest argument I have ever heard. Let me make sure
Iunderstand...your saying that because most hardware that Linux can run on is
too
weak to run Oracle that therefore Linux is too weak to run Oracle. Correct?
OK,
whatever...
You know as a whole the majority of production automobiles cannot go faster
than
about 120 MPH (guess, just an example), therefore, according to your logic no
production automobiles can go faster than 120 MPH. Wow, I better go tell all
those
people that have Corvette's, Ferrari's, Porsche's, etc. that they really
can't go that
fast, it is just a figment of their imagination...
--
Jim Zubb
ji...@ecom.net
Gerhard Wesser
In article <m2yb1ep...@lucifer.guardian.no>,
bo...@guardian.no (Bjorn Borud) wrote:
----------SCHNIPP----SCHNAPP-----------CUT-----------------------------
> do we know anyone who knows Larry personally and could give him a call
> and explain to him what Linux is and that there are some people who
> would really like to run Oracle on it? :-)
I have heard that Oracle from SCO should running under Linux (Bin-Emul.)
Is'nt ?
>
>
> PS: I'd like to run Designer/2000 on Linux as well :-)
I would like to get them too...xmas is soon..i will put my shoes outside
the door....maybe that helps :-)
>
> -Bjørn
> --
> Bjørn Borud <bo...@guardian.no> | "The Net interprets censorship
> <URL:http://www.pvv.unit.no/~borud/> | as damage and routes around it."
> UNIX person, one of "them" | - John Gilmore
--
Gerhard Wesser
dd...@linux.prima.no|spam.de http://www.prima.de
Jerry Gitomer
Tip,
excuse me, but where can I buy a two year old PC based system so I
won't be able to run Oracle on Linux?
By the way, I just checked and according to the CD-ROM I can run
Oracle on a 64MB Pentium with NT 4.0. Since NT 4.0 is bigger than
Linux I should certainly be able to run Oracle on a 64MB Linux
box. (Okay, so it won't be the fastest thing on earth -- but for
a single user it will suffice.)
A Linux based Pentium with 1GB of RAM should be able to comfortably
support a small to medium size organization.
Jerry
tip wrote (in part):
>
> i'm not talking about the later model pci chipset motherboards in the
> past two years, rather as a whole.
>
> as a whole, most motherboards that support linux do not support over
> 128 megs.
>
George Henry C. Daswani
tip (t...@blahblah.com) wrote:
: there is NO WAY the pc hardware end of linux would be able to handle running
: oracle in any decent manner. plus its tweaking is rather limited compared to
: solaris and hpux. i have run oracle on both suns and hp boxes - there's just
: no way linux is ready (yet) for oracle.
Oh shit, so I guess SCO, Solaris X86, NT can run them then right?
If you are gonna spew some *$(@& #$, might as well compare it to the
other platforms on X86..
Considering that Linux takes the least amount of resources compared to
SCO, Solaris X86, and NT.. It would be one of the best platforms to
run Oracle under X86.. I mean, Linux is quite a nifty threaded os which
beats out most others in context switching, memory footprint and etc..
: i currently run oracle 7.2.3 on hp9000/s800/k210's running hpux 10.10 with a
: gig of memory each. that's alright for running oracle.
: but imagine a pc with linux and a motherboard limitation of 128M, a 200Mhz
: processor... i don't think so...
Well, if you buy shit hardware then... I do believe FTP.CDROM.COM is running
of a box that has more than 512 Megs of RAM and it's X86 based...
I guess those #*(@&4 X86 boxes are good for nothing then considering
that ftp.cdrom.com (the biggest ftp server on the net) is running on the
X86 platform and turning out gigabytes of IO everyday albeit running
FreeBSD (your argument does not make sense bub).
If you buy SHIT PC hardware, then you get SHIT basically. Some new
X86 boxes are coming out with I20 tech (I believe DELL has poweredge
servers using em) and should alleviate some of the IO problems in general
pc design..
: linux is good for being a workstation, or a small server - but the heavy duty
: shit - better leave it to superior hardware and os's.
Quite funny.. How does one define heavy duty?
You should ask Nasa why they picked the X86/Linux platform running clustered
under the Beowulf project - price/performance ratio.
I do have a few gripes about using Linux for RDBMS because of
the 2 gig file limitation limit (which I hope will be fixed sooner or later).
But I do believe with the general move to clustering in 3.0, fine
grained locks for SMP, and future processors like the merced..
Linux should be able to DISH out TPC's without any problem in the future..
And besides, you can RUN Linux on an Alpha, PowerPC, and Sparc....
I do believe that SOLIDTECH (Solid SQL) and Software AG (Adabas D) will
make money in the future when IS managers finally realize that Linux gives
them so much for so little. I'm pretty sure Linux boxes are now sneaking
up behind the back doors onto most IS depts acting as web servers,
name servers, dhcp servers and etc. The day of reckoning will come
when one of the RDBMS heavy weights will deliver a shattering announcement
regarding porting their XYZ RDBMS to the Linux platform and actually
support it directly (But alas, Software AG only supports Linux indirectly).
I do believe that Oracle, Sybase, and Informix should losen up with
their restrictions and pricing. I believe they will finally realize
their mistake not porting to Linux when one day, microsoft is knocking
at their door carrying Windows NT 5.0, a good enough version of Microsoft
SQL server, and price/performance ratio that makes them look silly
when ran on Intel's merced processor.
(microsoft never makes outstanding software, but they do make software
that is good enough for most people)..
John S. Dyson
In article <67nud3$sma$2...@holocron.odc.net>,
gdas...@biddersparadise.com (George Henry C. Daswani) writes:
>
> Well, if you buy shit hardware then... I do believe FTP.CDROM.COM is running
> of a box that has more than 512 Megs of RAM and it's X86 based...
>
> I guess those #*(@&4 X86 boxes are good for nothing then considering
> that ftp.cdrom.com (the biggest ftp server on the net) is running on the
> X86 platform and turning out gigabytes of IO everyday albeit running
> FreeBSD (your argument does not make sense bub).
>
had a bad RAM bank.) Single processor/233 MHz. I think that we all sometimes
forget how much raw CPU power that we really have, most of it normally
being sucked up by GUI. In dhrystones (kind-of bogus), a dual P6 has approx
500X a VAX-11/780 or DEC-10 in CPU (Assuming a DEC-10 is a little faster
than an original VAX.)
Additionally, it is very easy to underestimate the amount of I/O that
a PC can do.
One should not expect a PC without multiple fans, running IDEs to work
reliably in that environment. Geesh, that machine appears to go through
SCSI drives also (I don't know how many have failed in the last few
years, but I wouldn't be surprised if it is 5 or more.) One problem with
building your own PC, is that it is very hard to do a component qualification
on such a small scale. Anecdotal information on the net is one way that
you can find out how reliable equipment is, except that the lead time
for that information is long. (Other people often start seeing problems
at the same time that you do on your deployed system.)
PC's are missing some features that some people want. Generally, they
aren't fault tolerant, and also not generally hot-pluggable (of course,
that can be arranged.)
James Youngman
>>>>> "tip" == tip <t...@blahblah.com> writes:
tip> In article <349fcd70....@news.mindspring.com>,
tip> Voytek Jarnot <vja...@mindspring.com> wrote:
>>
>> You are so full of it that it hurts me to waste the time to
>> reply. What exactly does 'most' mean to you??? Here's a
>> breakdown of some of intel's various chipsets, its the chipset
>> that the motherboard incorporates, not the motherboard itself,
>> that limits the amount of RAM.
tip> are you blind or just dumb? what does linux run on? 386, 486,
tip> pentium, whatever.
tip> i'm not talking about the later model pci chipset motherboards
tip> in the past two years, rather as a whole.
tip> as a whole, most motherboards that support linux do not support
tip> over 128 megs.
Yes, but nobody is expecting to run large Oracle servers on a 386.
The point is that well-specified new x86 machines *can* handle Oracle.
Michael Hasenstein
How about another reason: ~40% market share of Oracle DB (don't know
exactly, but it was a lot), i.e. many people have it already and it would
be extremely helpful to be able to connect a Linux box to an Oracle DB.
You can do this now, but only with workarounds. It would be much nicer if
one could use such things like OCI directly on Linux, which would be a
side effect of having Oracle ported to the platform.
Also, if your company already runs an Oracle DB (see market share, so this
is quite probable), wouldn't it be better to use the same system instead
of adding a different one? Ok, some people like to use as many different
products as possible in the same environment, but I guess that's not the
majority.
--
**************************************************************
Michael Hasenstein; Siemens Nixdorf (Paderborn, Germany)
currently @Oracle, Redwood Shores, California
http://www.csn.tu-chemnitz.de/home/mha/
Benjamin Redelings I
tip wrote:
>
> i'm not talking about the later model pci chipset motherboards
> in the
> past two years, rather as a whole.
Why?
> as a whole, most motherboards that support linux do not support over
> 128 megs.
OK, but so what? WHo cares about "most"? That's basically penalizing
Intel for having both high-end and low-end equipment available. Who
cares if there are also a lot of low-end machines... that's a different
market niche. Its mostly irrelevant to the issue of high-end machines.
The "average intel machine" is a meaningless, or at least useless,
construct. (sigh) The dangers of reification... The only time I can
see anyone want to talk like that is correcting some idiot manager who
has beliefs about an "average intel machine".
-BenRI
Andre Couture
ps; look below!!!
Voytek Jarnot wrote:
On 22 Dec 1997 20:59:11 GMT, tip <t...@blahblah.com> wrote:
>In article <349fcd70....@news.mindspring.com>,
>Voytek Jarnot <vja...@mindspring.com> wrote:
>>
>> You are so full of it that it hurts me to waste the time to reply.
>>What exactly does 'most' mean to you??? Here's a breakdown of some of
>>intel's various chipsets, its the chipset that the motherboard
>>incorporates, not the motherboard itself, that limits the amount of
>>RAM.
>
>are you blind or just dumb? what does linux run on? 386, 486, pentium,
>whatever.
>
>i'm not talking about the later model pci chipset motherboards in the
>past two years, rather as a whole.
>
>as a whole, most motherboards that support linux do not support over
>128 megs.
tip, Funny, I have a Tyan motherboard Dual PPro with 160Megabytes of RAM and Linux run just fine!!!Don't know where you get your info!!!
Maybe you ought to take a look at the topic under which you are
posting ... or maybe you did and are just too stypid to realize that
running an Oracle server on a 3/486 is not a viable option. I include
shipsets for those processors that are viable (though maybe not
optimal) options : Pentium, Pentium Pro, Pentium II.
right VJ, Oracle is a PIG, and it will be even on Linux!!
/Andre
--
Andre Couture
938934 Ontario Inc.
mailto:acou...@search-n-find.com
http://www.search-n-find.com/
A.van.Kessel
The real reason is - of course - stratigic.
Doing a linux-port would cause too many people
becoming an 'O' ## - expert. That would cost O-corporation
a lot of their revenues on consultancy. That would put al lot of O-suits
into unemployment, and severely damage the company's reputation.
BTW. don't overestimate the O-product. It is less solid than linux
(if this kind of comparison can be made between an OS and a DMBS *can* be made)
and is extremely version-sensitive (probably to keep consultancy and
retraining revenues high), which will cause "the" linux port to be
repeated every few months
## Note. Beeing an Ingres-DBA, I don't write the O-word in full. It can cause
disaster :-)
Happy Hacking,
--
Adriaan van Kessel.
Ingres DBA, C/Unix hacker
Email: Adriaan.v...@NotThere.rivm.nl
(remove NotThere. from the above address)
Eric Lee Green
On 21 Dec 1997 19:42:47 GMT, tip <t...@blahblah.com> wrote:
>there is NO WAY the pc hardware end of linux would be able to handle running
>oracle in any decent manner. plus its tweaking is rather limited compared to
>solaris and hpux. i have run oracle on both suns and hp boxes - there's just
Well, the latest SMP servers from Compaq and DEC look pretty
good. RAID-5 SCSI, hot-swappable drives, multiply-redundant
hot-swappable power supplies, what more could you want? The only thing
that will down it is if the motherboard or CPU burns up (no recent OS,
alas, is as reliable as the old Honeywell Multics, where a single
processor going out just meant disabling that processor and not even
having processes burp). I've seen one configured with 512mb of memory.
Now, true, this isn't going to run a large business, but it will
certainly run a small-to-midsize business without even breaking a
sweat.
>no way linux is ready (yet) for oracle.
Well, I'd agree, to a certain extent. Oracle wants to do like they do with
SCO -- put out a version and have it still running 5 years later without
even a recompile. The Linux world is famous for shifting hither and fro.
Heck, there's not even a stable libc yet.
But none of this has anything to do with whether Oracle can technically be
run under Linux. The only current limitation is that it'd have to use raw
partitions for its disk storage, since ext2fs is limited to 2gb for file size.
>i currently run oracle 7.2.3 on hp9000/s800/k210's running hpux 10.10 with a
>gig of memory each. that's alright for running oracle.
>
>but imagine a pc with linux and a motherboard limitation of 128M, a 200Mhz
>processor... i don't think so...
Err, obviously you haven't checked out the "big iron" PC servers used
for SCO Unix and Windows NT. Most of them don't SHIP with less than
128m of memory, much less have a LIMIT of 128m. In fact, SCO had to
put out an OpenServer revision because they didn't properly support
1gb of memory (only 512mb) and somebody caught them on it. 200mhz
processors? Well, the quad-Pentium Pro, certainly. But a quad Pentium
Pro box has more CPU horsepower than the fancy hp9000/etc. boxes, it's
the I/O that has traditionally been the bottleneck of PC-family
computers. Still is, for that matter. But the I/O bottleneck has
narrowed greatly with hardware RAID controllers with bus-mastering
PCI.
>linux is good for being a workstation, or a small server - but the heavy duty
>shit - better leave it to superior hardware and os's.
Well, Linux also runs on that superior hardware :-). (See: Linux/SPARC,
Linux/ALPHA, Linux/PowerPC). In fact, Linus Torvalds' only Intel-based PC
is an old '386 he keeps around for nostalgic purposes -- he does all his
own work on an Alpha running Red Hat Linux.
As for the "superior OS" bit, the biggest limitations in Linux are in
the file system. The ext2fs is a good simple clean little file system,
but not exactly featuresome. There's nothing stopping anybody from
improving ext2 (create ext3?) or implementing a new file system --
Linux has supported multiple filesystems with ease since the Linux 1.1
days -- but nobody has done it. There's also various clunkiness at the
OS level, but let's face it, all versions of Unix have accreted a bit
of clunkiness here and there. Heck, there's still stuff in Unix that
exists for no reason other than that K&R&P did it that way back in
1975.
Somebody brought up "well, it's not as tunable". Most of the tuning decisions
of the past, such as, e.g., how many buffers to allocate for i/o operations,
have been taken over by the dynamic buffer cache. A dynamic buffer cache is
provably optimal, since it is always in sync with your usage patterns, while
a statically-allocated buffer cache must be tuned every time you change your
work load or usage patterns. The algorithms used for the dynamic buffer cache
probably need a bit of tuning themselves, but there's nothing at all wrong
with the idea of a dynamic buffer cache.
(And do note that there's still a LOT you can tune about Linux -- it's just
that, unlike with older Unixes, the default Linux values tend to be reasonable
enough that nobody ever bothers).
--
Eric Lee Green ex...@softdisk.com Executive Consultants
Systems Specialist Educational Administration Solutions
You might be a redneck if you put on insect repellant prior to a date.
Joel Garry
In article <67nud3$sma$2...@holocron.odc.net>,
George Henry C. Daswani <gdas...@biddersparadise.com> wrote:
>
>I do believe that Oracle, Sybase, and Informix should losen up with
>their restrictions and pricing. I believe they will finally realize
>their mistake not porting to Linux when one day, microsoft is knocking
>at their door carrying Windows NT 5.0, a good enough version of Microsoft
>SQL server, and price/performance ratio that makes them look silly
>when ran on Intel's merced processor.
That would be nice, but they have stockholders to answer to (Larry just
lost $2G one day...). Watch for a siege mentality as they try to keep their
profitability up, much like DEC, WANG, etal in the 80's...
>
>(microsoft never makes outstanding software, but they do make software
>that is good enough for most people)..
And leaves many people outstanding in the cold...
Russell Nelson
tip <t...@blahblah.com> writes:
> there is NO WAY the pc hardware end of linux would be able to handle running
> oracle in any decent manner. plus its tweaking is rather limited compared to
> solaris and hpux. i have run oracle on both suns and hp boxes - there's just
> no way linux is ready (yet) for oracle.
>
> i currently run oracle 7.2.3 on hp9000/s800/k210's running hpux 10.10 with a
> gig of memory each. that's alright for running oracle.
>
> but imagine a pc with linux and a motherboard limitation of 128M, a 200Mhz
> processor... i don't think so...
So... you're saying that Oracle doesn't run on SCO or Solaris/x86 either.
It's the same hardware, you know.
--
-russ <nel...@crynwr.com> http://www.crynwr.com/~nelson | Freedom is the
Crynwr Software supports freed software | PGPok | primary cause of peace.
521 Pleasant Valley Rd. | +1 315 268 1925 voice | Obedient, Christian, statist:
Potsdam, NY 13676-3213 | +1 315 268 9201 FAX | you only get to pick two.
Leigh Porter
tip wrote:
> there is NO WAY the pc hardware end of linux would be able to handle running
> oracle in any decent manner. plus its tweaking is rather limited compared to
> solaris and hpux. i have run oracle on both suns and hp boxes - there's just
> no way linux is ready (yet) for oracle.
>
> i currently run oracle 7.2.3 on hp9000/s800/k210's running hpux 10.10 with a
> gig of memory each. that's alright for running oracle.
>
> but imagine a pc with linux and a motherboard limitation of 128M, a 200Mhz
> processor... i don't think so...
>
> linux is good for being a workstation, or a small server - but the heavy duty
> shit - better leave it to superior hardware and os's.
>
> thanks,
> - tip
Superious hardware, maybe - Superious OS - Errm, nope! Solaris on this Ultra1
has crashed
and ran out of memory more times than any of my Linux boxes have!
Now a Linux port to some big hardware, oh that's already done - Just need Oracle
to run on
Linux on a SUN/whatever and your away :)
--
Leigh
Leigh Porter
Thomas Schenk wrote:
>
>
> This is total crap. The entire Deja News service is run on Linux
> machines
> and Deja News has one of the largest databases on the Web. I also don't
> know what decade you are living in, but we have Linux SMP running on
Hey, would the DejaNews people - or have they written some gumph on how they set
it up and what hardware it runs on so that the world can be told?
That would be pretty cool, a nice big case study!
--
Leigh
Leigh Porter
tip wrote:
> In article <349fcd70....@news.mindspring.com>,
> Voytek Jarnot <vja...@mindspring.com> wrote:
> >
> > You are so full of it that it hurts me to waste the time to reply.
> >What exactly does 'most' mean to you??? Here's a breakdown of some of
> >intel's various chipsets, its the chipset that the motherboard
> >incorporates, not the motherboard itself, that limits the amount of
> >RAM.
>
> are you blind or just dumb? what does linux run on? 386, 486, pentium,
> whatever.
>
> i'm not talking about the later model pci chipset motherboards in the
> past two years, rather as a whole.
>
> as a whole, most motherboards that support linux do not support over
> 128 megs.
Is there a motherboard that does not support Linux and supports more than
128Mb?
--
Leigh
Kristian Koehntopp
e_l_...@hotmail.com (Eric Lee Green) writes:
>Well, the latest SMP servers from Compaq and DEC look pretty
>good. RAID-5 SCSI, hot-swappable drives, multiply-redundant
>hot-swappable power supplies, what more could you want?
Okay, since you asked - how about these:
Hot swappable CPUs, hot swappable memory modules, hot swappable
mainboards, individual temperature control per CPU, individual
CPU reboot without service disruption, the ability to "farm"
processors and run multiple releases of the operating system on
the same iron (one OS per "farm") and to dynamically reassign
processors from one "farm" to another without service
disruption, a crossbar-switched backplane instead of a bus
system, two pathes to every piece of hardware and an operating
system that knows how to reroute data pathes within the machine,
fibre channel arbitrated loop (FC-AL, 1 Gigabit/sec and up to 10
km distance between disk array and machine) to connect to the
disk arrays (using RAID 1+0 instead of RAID 5 for speed), up to
14, 28 or even 64 processors per machine (64 bit, 250 MHz, 4 MB
cache per processor for example) and up to 14, 28 or even 64 GB
main memory per machine. Then two machines of these with a
working clustering/failover-solution.
I admit that you can get pretty decent boxes when looking for PC
servers (We are using IBM 704's as standard servers and are
quite satisfied), but the above features are usually missing in
PC hardware - even server hardware.
That's what you get when you buy Enterprise Ultra hardware from
Sun. This is at least one league above PC hardware (yes, in
features and in price as well).
You don't install such boxes routinely, but for a data warehouse
solution or other mission-critical parts of your enterprise this
is just right and it scales like nothing (from single processor
solutions with memory in the 32 MB range up to 64 processor
solutions with GBs of mem and TBs of disk space).
>The only thing
>that will down it is if the motherboard or CPU burns up (no recent OS,
>alas, is as reliable as the old Honeywell Multics, where a single
>processor going out just meant disabling that processor and not even
>having processes burp).
Suns Enterprise Ultra Series and Solaris do that, too.
Kristian
--
Kristian Koehntopp, http://www.koehntopp.de
"3*666 = 1998" -- Bruder 13 auf der Verschwoerungsmailingliste
David A Davidson
Years ago Progress touted of its cross platform capabilities. I think Progress is heading away from the Unix world of open systems to the closed, proprietary
systems of the windows world. That started when Progress no longer supported
the motif environment. So much for client/server expansion?
--
David A Davidson <ddav...@netcom.ca>
HAM RADIO - VA3DAD <va3...@unix.ve3uow.uwaterloo.ca>
Bill Beaton
tip wrote:
> In article <349fcd70....@news.mindspring.com>,
> Voytek Jarnot <vja...@mindspring.com> wrote:
> >
> > You are so full of it that it hurts me to waste the time to reply.
> >What exactly does 'most' mean to you??? Here's a breakdown of some of
> >intel's various chipsets, its the chipset that the motherboard
> >incorporates, not the motherboard itself, that limits the amount of
> >RAM.
>
> are you blind or just dumb? what does linux run on? 386, 486, pentium,
> whatever.
>
You really aren't with it, are you! I run at home on my dual PPro (using the
SCO Oracle), and I've installed it on Sparcs and Alphas.
> i'm not talking about the later model pci chipset motherboards in the
> past two years, rather as a whole.
>
> as a whole, most motherboards that support linux do not support over
> 128 megs.
>
Really? Then why would a lot of people run ASUS MBs, which support 512MB to
1GB? Before you get into a technical discussion, at least have a minimal
knowledge of the technology available, and don't spew utter shit.
Bill Beaton
Nix
In article <D5864FBF358B6707.FC73AFE2...@library-proxy.airnews.net>,
Eric Lee Green <e_l_...@hotmail.com> wrote:
> A dynamic buffer cache is
> provably optimal, since it is always in sync with your usage patterns, while
> a statically-allocated buffer cache must be tuned every time you change your
> work load or usage patterns. The algorithms used for the dynamic buffer cache
> probably need a bit of tuning themselves
I'll say! My Linux box slowed down drastically after 2.0.29, because the
buffer caching code was rewritten. Before the change, the buffer cache on my
8Mb system was around 200-500Kb; now, it's 2.5Mb or higher. This means that
with the kernel X running too, I've got less than 800Kb free for other
processes. It really would be *much* faster if I could set a forced maximum
size or something; I've looked at the code to try to do that but it's such
a tangle I hardly know where to start. :(
--
"Computers in the future may weigh no more than 1.5 tons."
-- Popular Mechanics, forecasting the relentless march of
science, 1949
Vlad Petersen
Nix wrote:
...........
> I'll say! My Linux box slowed down drastically after 2.0.29, because the
> buffer caching code was rewritten. Before the change, the buffer cache on my
> 8Mb system was around 200-500Kb; now, it's 2.5Mb or higher. This means that
> with the kernel X running too, I've got less than 800Kb free for other
> processes. It really would be *much* faster if I could set a forced maximum
> size or something; I've looked at the code to try to do that but it's such
> a tangle I hardly know where to start. :(
Try something like this:
echo "128 256 1024" > /proc/sys/vm/freepages
It may not be exactly what you want but it helps. Experiment
with the values a little bit and see if it makes a difference.
The above was optimal for my system with 48MB of RAM and 48MB
of swap partition. I think I saw it explained in some linux*
newsgroups in more details, try to find it with Dejanews.
As far as I understood, the buffer caching algorithm rewrite
you mentioned didn't really make Linux slower: its defaults
only made the performance curve flatter which is good.
--
Vlad Petersen | <vladimip at uniserve dot com>
#include <disclaimer.h> | *Good pings come in small packets*
Vancouver, BC | Windows: for IQs smaller than 95
SIGSIG -- signature too long (core dumped)
Kristian Koehntopp
Leigh Porter <le...@wisper.net> writes:
>Hey, would the DejaNews people - or have they written some gumph on how they set
>it up and what hardware it runs on so that the world can be told?
--- quote on ---
~From: tsc...@dejanews.com (Thomas Schenk)
~Newsgroups: netuse.lists.linux-smp
~Subject: Re: How stable and useful are SMP Linux now?
~Date: 3 Dec 1997 22:31:02 +0100
On 03-Dec-97 Dave Andruczyk wrote:
>>
>> Our lab is going to buy new server and use Linux on it.
>> We are thinking about two-processor Pentium II machine.
>> (maybe it really become 2-cpu later, but smp-motherboart just
>> now).
>> Does SMP Linux enough usable for production machine now?
>
> I manage server dual Pentium Pro servers, and a Single PII server. All
> of them are rock solid under linux. The PPro are running on a FIC PN6210
> mainboard, 128 megs ram, aha2940UW scsi, with 256k cache chips. The P11
> is running off of a Tyan Tiger Mboard wiht 128 megs ram. All systems see
> typically high loads, sometimes over 20 ( load avg), and have never
> failed due to the OS.
>
>> Will it be so where 2.2.xx will be released?
>
> PII's are great with only 512 megs of RAM, Ppro's can handle 1Gig.
>
That's nothing. ;) The entire Deja News search engine runs on Linux SMP
systems (>50 dual PPro currently and moving to dual Pentium IIs). These
machines handle over 3.5 Million hits per day, with peak load of about
50 page views per second. They have already ditched one operating system
for instability and if Linux SMP wasn't reasonably stable, they would have
ditched it long ago. The most common cause for a machine failure is power
outages. We have seen some instability in 2.0.31, but for our hardware,
2.0.32 is pretty solid.
Tom Schenk
System Admin
---
+-----------------------+-------------------------+------------------------+
| Tom Schenk | Use Linux! | All opinions expressed |
| tsc...@dejanews.com | Friends don't let | are mine and not those |
| tsc...@theoffice.net | friends do Windows! | of my employer. |
+-----------------------+-------------------------+------------------------+
--- quote off ---
graywolf
Have any of you guys' priced Oracle? Last time I checked it was
$8500 (pc version --you don't even want to know about the unix
version).i
--
Graywolf
(Tom Rittenhouse)
---------------------------------------------------------------
A James Lewis
Kevin Forge wrote:
>
> Leigh Porter wrote:
>
> > tip wrote:
> >
> > > In article <349fcd70....@news.mindspring.com>,
> > > Voytek Jarnot <vja...@mindspring.com> wrote:
> > > >
> > > > You are so full of it that it hurts me to waste the time to reply.
> > > >What exactly does 'most' mean to you??? Here's a breakdown of some of
> > > >intel's various chipsets, its the chipset that the motherboard
> > > >incorporates, not the motherboard itself, that limits the amount of
> > > >RAM.
> > >
> > > are you blind or just dumb? what does linux run on? 386, 486, pentium,
> > > whatever.
> > >
> > > past two years, rather as a whole.
> > >
> > > as a whole, most motherboards that support linux do not support over
> > > 128 megs.
> >
> > 128Mb?
> >
> > --
> > Leigh
>
The idea that there might be a memory limitation never even occured to
me, where did these guys get the idea that either there's an x86
motherboard that cant run Linux? I have x86 boards that wont run
WINDOWS! OK, it's because it cant autodetect things on startup....
Still, I'm running linux on them.... the one I'm typing on supports up
to 768Megs.....
James
Kevin Forge
Ph
tip <t...@blahblah.com> wrote in article <67jrfn$h...@eve.enteract.com>...
> there is NO WAY the pc hardware end of linux would be able to handle
running
> oracle in any decent manner. plus its tweaking is rather limited compared
to
> solaris and hpux. i have run oracle on both suns and hp boxes - there's
just
> no way linux is ready (yet) for oracle.
>
> i currently run oracle 7.2.3 on hp9000/s800/k210's running hpux 10.10
with a
> gig of memory each. that's alright for running oracle.
>
> but imagine a pc with linux and a motherboard limitation of 128M, a
200Mhz
> processor... i don't think so...
>
How many intel/ Win NT's are running with 64 MB + Oracle ? You would be
surprised , so for small database needs Linux will do (and Linux
performance is generally better than NT's). I know of hundreds of systems
running Oracle with less than 128MB RAM on HPUX, Digital Unix, AIX, NT ...
and they run in production too.
IMO, Oracle does not need to port much to support Linux - I know some of
their employees ran Oracle on Linux at home
Some years ago, I ran Oracle 7.0 on Linux with IBCS just as for a test on a
486 with just 16MB Ram -- that is sixteen megabyte !!!! and (to my suprise)
it worked - I agree that no business would want to run on such a platform,
but for learning Oracle in a Unix environment it was great.
- Phil.
Stephen Harris
Kristian Koehntopp (kr...@koehntopp.de) wrote:
[ Spec deleted ]
: That's what you get when you buy Enterprise Ultra hardware from
: Sun. This is at least one league above PC hardware (yes, in
: features and in price as well).
Errm, your spec is actually the Ultra 10000 series. The U3000 -> U6000
series don't even come close to this in most aspects. But since we are
talking well hundreds of thousands of UK pounds for such a spec, I really must
assume a comment such as yours is merely a troll, not meant for logical
discussion. Even a U3000, dual CPU, 1Gb RAM, dual SSA's have a list price
of 100,000 pounds and this is pennies compared to an U10000.
Put a PC against an Ultra 2 series machine, then we get useful comparisons.
I don't like PC's much (though they make good desktop machines, and
workgroup servers), but compare like to like!
The U10000 machine is getting close to "big iron" systems - the closest of
any Unix machines around at present.
PS, my Ppro200, 64Mb RAM, EIDE disk running SCO Oracle can select and
insert a million rows in under 4 minutes "insert into x select * from x"
which is comparible to the results the above U3000 system gave. I think
I damaged the hard disk though through excessive seeking :-)
--
Stephen Harris
sw...@spuddy.mew.co.uk http://www.spuddy.org/
The truth is the truth, and opinion just opinion. But what is what?
* Meeeeow ! Call Spud the Cat on > 01268 515441 < for free Usenet access *
Ben Lindstrom
It's funny.. UNIFY 3.0 (I know it's OLDER then the hills =) running
on Solaris 2.5.1 on Dual P2-200 is what most the US Courts are moving
to. Only 64megs of memory (could be 128meg...Don't remember off hand).
About 40 people docketing (at least at the smaller branches) and it
works like a charm. Unless one is building the *NEXT* Airline
system or expecting a shitload of clients at one time. P2-200 would
work just fine...
If Solaris can deal with UNIFY 3.0 (It's even doing AXP emulation!)
then Linux should beable to handle Oracle, Informix (SE at least), or
Sybase.
In comp.os.linux.development.apps Jerry Gitomer <jgit...@p3.net> wrote:
> Tip,
> excuse me, but where can I buy a two year old PC based system so I
> won't be able to run Oracle on Linux?
> By the way, I just checked and according to the CD-ROM I can run
> Oracle on a 64MB Pentium with NT 4.0. Since NT 4.0 is bigger than
> Linux I should certainly be able to run Oracle on a 64MB Linux
> box. (Okay, so it won't be the fastest thing on earth -- but for
> a single user it will suffice.)
> A Linux based Pentium with 1GB of RAM should be able to comfortably
> support a small to medium size organization.
[...snip..]
Alex Buell
On 18 Dec 1997, Christopher B. Browne wrote:
> Give threading six months to solidify under the GLIBC "regime," and we can
> provide a somewhat different story...
I keep hearing about GLIBC. What's the difference between that and a
normal LIBC?
My employers uses mSQL which, interestingly enough runs on Linux
platforms, so I can recompile their big UN*X (Sun/HP/whatever)
applications on this here 486DX/120 with Linux and have it doing precisely
the same functions as on a big kick-ass Sparc Ultra.
Now that's portability!
Try doing that for Windows, but that's a different story and only lives on
Intel and clones chips!
Cheers,
Alex
--
/\_/\ Make the police happy,
( o.o ) Smoke some cannabis today!
> ^ < Peace, Love, Unity and Respect to all.
http://www.tahallah.demon.co.uk
Arun Sharma
Alex Buell <al...@tahallah.demon.co.uk> writes:
> On 18 Dec 1997, Christopher B. Browne wrote:
>
> > Give threading six months to solidify under the GLIBC "regime," and we can
> > provide a somewhat different story...
>
> I keep hearing about GLIBC. What's the difference between that and a
> normal LIBC?
Builtin multithreading (pthreads) and thread-safe implementation of
libc.
-Arun
Chris Zimman
> >Well, the latest SMP servers from Compaq and DEC look pretty
> >good. RAID-5 SCSI, hot-swappable drives, multiply-redundant
> >hot-swappable power supplies, what more could you want?
>
> Okay, since you asked - how about these:
>
* spurious garbage deleted *
> I admit that you can get pretty decent boxes when looking for PC
> servers (We are using IBM 704's as standard servers and are
> quite satisfied), but the above features are usually missing in
> PC hardware - even server hardware.
>
> That's what you get when you buy Enterprise Ultra hardware from
> Sun. This is at least one league above PC hardware (yes, in
> features and in price as well).
>
> You don't install such boxes routinely, but for a data warehouse
> solution or other mission-critical parts of your enterprise this
> is just right and it scales like nothing (from single processor
> solutions with memory in the 32 MB range up to 64 processor
> solutions with GBs of mem and TBs of disk space).
This is the same argument I've heard a thousand times over. More so
than a technology issue, this is an elitist issue. It seems that for
some reason unbeknown to me, some admins (especially) and coders are
'threatened' by the idea of having something similar in power to what
is available to the home user these days.
This seems especially true when it comes to free OSs versus commercial
ones. The quick rate of development of free OSs takes a lot of the
bragging rights away from some people. I'm sure that we've each
encountered some of them.
Typically, in my experiences, they will make a huge bravado of
hardware that's bleeding edge and so expensive that it can only be
afforded by the largest of corporations, it also is usually
extensively proprietary. This is often followed by a "...and that's
why PCs and free OSs will never be useful in 'real business'
situations...but that's not to say they're not good machines with a
purpose, etc."
I've heard this a thousand times over, and I laugh everytime. I
always remember that these were the same things that IBM and Digital
told their customers as PCs were replacing 3270 terminals. "...Oh,
PCs...pay them no mind, they're for home users...they'll never be used
in business environments..."
I certainly don't need to explain the rest.
Granted, there is some truth in the redundancy of larger 'iron'
machines. But, just as it happened before, this is going to become
less and less of an issue. In many cases with PCs, it's equivalent or
cheaper to replace an entire PC than to replace a single failed
component on a bigger iron machine. Ever look at vendor prices for
replacement drives, power supplies, etc?
Some companies and people like to pretend that if they quietly ignore
and or downplay PCs are free operating systems, they'll just dry up
and disappear. They are free to revel in their ignorance.
If Oracle makes a decided point NOT to port to Linux or to be
generally elitist to appeal to assinine admins, so be it, they'll be
the ones missing out on a massing market.
--Chris
Nilanjan Bhowmik
>
> Granted, there is some truth in the redundancy of larger 'iron'
> machines. But, just as it happened before, this is going to become
> less and less of an issue. In many cases with PCs, it's equivalent or
> cheaper to replace an entire PC than to replace a single failed
> component on a bigger iron machine. Ever look at vendor prices for
> replacement drives, power supplies, etc?
This may not be the right context but I can't help flame SUN a little.
SUN is touting their big Iron killer machines. They are the main
supplier of software + hardware to ensure smooth operation and this
gives them a big edge to provide reliable operation. I bought six
Trunking capable Quad Ethernet Cards a month back ( paid a lot for
these things ).
When I tried to install Solaris 2.6 + Oracle etc. etc. I found out that
the card is not being recognised !! I went through lot of things before
I called SUN and found out THEY DO NOT HAVE DRIVER FOR THEIR OWN LATEST
HARDWARE RUNNING THEIR OWN LATEST OPERATING SYSTEM - SOLARIS 2.6 !!!
Have you ever tried to downgrade Solaris _after_ you have all the
software installed, tables done etc. etc. - don't even try ...
Anyway, My point is SUN is still miles away from catching the big Iron.
A company who can not synchronize their development cycle may not be
trusted with your bank account.
> If Oracle makes a decided point NOT to port to Linux or to be
> generally elitist to appeal to assinine admins, so be it, they'll be
> the ones missing out on a massing market.
>
Here is a good one - Oracle please please please, at least the Client
OCI part will be great !!
Nilanjan
Alex Buell
On 28 Dec 1997, Arun Sharma wrote:
> > I keep hearing about GLIBC. What's the difference between that and a
> > normal LIBC?
>
> Builtin multithreading (pthreads) and thread-safe implementation of
> libc.
Aha. Thanks.
Kristian Koehntopp
sw...@spuddy.mew.co.uk (Stephen Harris) writes:
>But since we are
>talking well hundreds of thousands of UK pounds for such a spec, I really must
>assume a comment such as yours is merely a troll, not meant for logical
>discussion.
I know very well that an Enterprise Ultra (be it the 3000 or the
10000) is nowhere near any PC server - not in features and not
in price. That is just the point I wanted to make. Customers
that absolutely need reliability, configurability, space to grow
or just raw performance won't go with any PC based server just
for the features I listed.
>The U10000 machine is getting close to "big iron" systems - the closest of
>any Unix machines around at present.
Close enough for many banking installations to seriously
consider to throw out their IBM hosts and install Sun hardware.
Kristian Koehntopp
Alex Buell <al...@tahallah.demon.co.uk> writes:
>On 18 Dec 1997, Christopher B. Browne wrote:
>> Give threading six months to solidify under the GLIBC "regime," and we can
>> provide a somewhat different story...
>normal LIBC?
glibc is the GNU C runtime library Version 2.0 (labelled Version
6 in Linux because of ELF version number requirements). Unlike
Linux libc Version 5 it is portable between differen Unices. It
also has much better thread support and many, many other
improvements over normal libc. RedHat and Debian are both deep
in the process of glib'ifiying their distributions and other
distribution vendors will follow.
The migration process will have a similar impact on Linux as the
transition from a.out to ELF (new library has many new features,
old applications will still require the old library installed
and loaded, selection between libraries is automatic, some
people will inevitable ruin their system when upgrading by
destroying critical system libraries and so on).
M Sweger
Stephen Harris (sw...@spuddy.mew.co.uk) wrote:
: Kristian Koehntopp (kr...@koehntopp.de) wrote:
: [ Spec deleted ]
: : That's what you get when you buy Enterprise Ultra hardware from
: : Sun. This is at least one league above PC hardware (yes, in
: : features and in price as well).
: Errm, your spec is actually the Ultra 10000 series. The U3000 -> U6000
: series don't even come close to this in most aspects. But since we are
: talking well hundreds of thousands of UK pounds for such a spec, I really must
: assume a comment such as yours is merely a troll, not meant for logical
: discussion. Even a U3000, dual CPU, 1Gb RAM, dual SSA's have a list price
: of 100,000 pounds and this is pennies compared to an U10000.
: Put a PC against an Ultra 2 series machine, then we get useful comparisons.
: I don't like PC's much (though they make good desktop machines, and
: workgroup servers), but compare like to like!
: The U10000 machine is getting close to "big iron" systems - the closest of
: any Unix machines around at present.
: PS, my Ppro200, 64Mb RAM, EIDE disk running SCO Oracle can select and
: insert a million rows in under 4 minutes "insert into x select * from x"
: which is comparible to the results the above U3000 system gave. I think
: I damaged the hard disk though through excessive seeking :-)
: --
: Stephen Harris
: sw...@spuddy.mew.co.uk http://www.spuddy.org/
: The truth is the truth, and opinion just opinion. But what is what?
: * Meeeeow ! Call Spud the Cat on > 01268 515441 < for free Usenet access *
Hi,
So how long does it take to read a million rows. I'm really interested
since I have such a need to know.
TIA
--
Mike,
mik...@whiterose.net
M Sweger
Kristian Koehntopp (kr...@koehntopp.de) wrote:
Hi,
So does this mean I still need the pthread library to write
applications using threads?
--
Mike,
mik...@whiterose.net
martin...@notexistent.org
Kristian Koehntopp <kr...@koehntopp.de> wrote:
[...]
> The migration process will have a similar impact on Linux as the
> transition from a.out to ELF (new library has many new features,
> old applications will still require the old library installed
Sorry, but in difference to libc4 vs. libc5 this doesn't work correct in any
case. You might have to be careful. If you get a libc5 binary of
ImageMagick's 'display', have a libc-5.4.38 and have already installed
libXpm, libjpeg, libpng, libtiff for glibc2, then 'display' will stop with
segfault or something similar.
Martin.
--
martin "at" quickstep "punkt" dirnet "punkt" com
--
Unix _IS_ user friendly - it's just selective about who its friends are !
--------------------------------------------------------------------------
Bill Eldridge
> > I admit that you can get pretty decent boxes when looking for PC
> > servers (We are using IBM 704's as standard servers and are
> > quite satisfied), but the above features are usually missing in
> > PC hardware - even server hardware.
> >
> > Sun. This is at least one league above PC hardware (yes, in
> > features and in price as well).
> This is the same argument I've heard a thousand times over. More so
> than a technology issue, this is an elitist issue. It seems that for
> some reason unbeknown to me, some admins (especially) and coders are
> 'threatened' by the idea of having something similar in power to what
> is available to the home user these days.
Ummm....I think he was talking about hardware differences,
which apparently why ATM works twice as fast running Linux
on Sparc vs. Intel (something about the interrupt structure),
why SGI's O2 has better graphics performance by bypassing
the PCI bus for direct memory access, etc.
I don't think anyone ever exclaimed that the Intel setup is
the most elegant in the world. It's gotten a lot faster than
before, but...
ben...@xs4all.nl
In comp.os.linux.development.system Kristian Koehntopp <kr...@koehntopp.de> wrote:
> The migration process will have a similar impact on Linux as the
> transition from a.out to ELF (new library has many new features,
aha, more disk space wasted, more memory wasted, slower system ? hmmpf. how
happy we are :-(.
> old applications will still require the old library installed
> and loaded, selection between libraries is automatic, some
> people will inevitable ruin their system when upgrading by
> destroying critical system libraries and so on).
so, the glibc can be installed together with the old version ? e.g. it's
installable on current systems ?
(how long do I need a.out and elf on the system hmmpf)
--
Grobbebol's Home | Don't give in to spammers.
http://www.xs4all.nl/~bengel | Use your real e-mail address
Linux 2.0.33 on an i586/64 MB | on Usenet.
martin...@notexistent.org
ben...@xs4all.nl wrote:
> aha, more disk space wasted, more memory wasted, slower system ? hmmpf. how
> happy we are :-(.
This is quite true.
> so, the glibc can be installed together with the old version ? e.g. it's
> installable on current systems ?
I do this for more than half a year and I am quite happy with it. You should
have at least a core system for glibc. After that, you can go the way to
migrate your system by compiling new software / new releases for glibc.
About 90 % of the software of my daily use is already converted. For some
pieces you need a small patch.
> (how long do I need a.out and elf on the system hmmpf)
That depends ....
Christopher Browne
On 30 Dec 1997 18:01:35 GMT, martin...@notexistent.org
<martin...@notexistent.org> wrote:
>> so, the glibc can be installed together with the old version ? e.g. it's
>> installable on current systems ?
>
>I do this for more than half a year and I am quite happy with it. You should
>have at least a core system for glibc. After that, you can go the way to
>migrate your system by compiling new software / new releases for glibc.
>About 90 % of the software of my daily use is already converted. For some
>pieces you need a small patch.
>
>> (how long do I need a.out and elf on the system hmmpf)
>
>That depends ....
Indeed.
If you use, as much as possible, software that is managed using
automated tools such as RPM or dpkg, then the vast majority of the
software should be able to be "upgraded" to use GLIBC, with some
hopeful minor performance improvements.
I write as many of my utilities as possible using scripting languages
like Perl, which has the natural effect that moving to GLIBC almost
goes unnoticed.
Things that are distributed as binaries will obviously require old
libraries. Which is typically commercial software...
--
"Bother," said Pooh, "Eeyore, ready two photon torpedoes and lock
phasers on the Heffalump, Piglet, meet me in transporter room three"
cbbr...@hex.net - "What have you contributed to Linux today?..."
<http://www.hex.net/~cbbrowne/lsf.html>
Stephen Harris
Kristian Koehntopp (kr...@koehntopp.de) wrote:
: I know very well that an Enterprise Ultra (be it the 3000 or the
: 10000) is nowhere near any PC server - not in features and not
: in price. That is just the point I wanted to make. Customers
: that absolutely need reliability, configurability, space to grow
: or just raw performance won't go with any PC based server just
: for the features I listed.
There is no argument here. It was made very early on in this debate. But
we are not comparing a 10,000 pound PC to a 500,000 pound server. Regardless
of the OS (Linux, NT, HPUX, Solaris) you would not consider a 10,000 piece
of hardware in an environment where you need a 500,000 pound piece of
hardware.
Argue like to like. Argue a Linux machine against a SCO Openserver machine
against an NT server against Sun Ultra 2 against... In all the other
environments I just mentioned, Oracle is supported and Linux is a comparable
platform.
Why not consider a Linux platform in an environment where SCO or NT would be
a valid platform? Why not have a Linux port of Oracle?
Kristian Koehntopp
ben...@xs4all.nl writes:
>so, the glibc can be installed together with the old version ? e.g. it's
>installable on current systems ?
It has a different major library number (libc.so.5 vs libc.so.6)
and the system will automatically select the appropriate libc
for every binary. Running a mixed system means having two libc's
in RAM, though.
Kristian Koehntopp
sw...@spuddy.mew.co.uk (Stephen Harris) writes:
>Why not consider a Linux platform in an environment where SCO or NT would be
>a valid platform? Why not have a Linux port of Oracle?
No problem here: Linux can certainly support a large database
system. At least it has all the features that are potentially
needed to do a port (or a new development) and the optimizations
that may be required for larger databases can easily be done.
Some remaining more fundamental problems (like large files and
large filesystems) are being worked on.
Two main problems with Linux remain, though. These are a) a very
heterogenous installed base (different distributions and
different releases of libraries) and b) a general unwillingness
to work with no-source-available type of packages.
Both Software AG and Star Division, who currently have larger
packages for sale (Adabas-D SQL Database and Star Office
Package) as native Linux versions, are currently facing these
problems and develop strategies to deal with them. For example,
both companies list exactly which distrbutions and versions of
distributions have been tested with their products (such as
"Suse Linux 5.0 or higher" or "Caldera OpenSomething x.y or
better"). But still a great deal of support is installation
support - shrink-wrapped installation as on other Unix platforms
or under Windows does not work as good.
Re b): People working with Star Office 4 (SO4) have reported a
certain slowness in startup, which is not visible when using the
same program release under Windows on the same machine. The
problem may be related to ld.so problems or with the large
number of shared libraries SO4 uses. Anyway, there has been a
certain hostility shown towards the Stardiv developers when they
discussed the problem in de.comp.os.linux.misc. True, the group
of people telling them to go away or release source has been
small, but overly noisy. But this is certainly not the way to
encourage porting to Linux.
Robert M. Stockmann
Michael Hasenstein wrote:
>
> Right.
> There was a sub-thread about traditional Unix-workstations vs Intel-WS,
> mostly saying that the Intel stuff (well, the components and the system
> architecture) is of (much?) lower quality. While this may be true for
> most or even all machines bought for low $ at the nearest computer shop
> I'd say it isn't for the big Intel servers sold by e.g. Compaq or
> Siemens or by (whoever) for running Windows NT (up to 8 processors by
> now, how NT scales (or not) is a different story). How many databases
> are there that need a 24 CPU server with 24 GB RAM and some TeraByte
> disk space? Of course, companies running these are our "most valuable"
> customers but guess why Oracle is available on NT, and I'd guess Linux
> has at least the same potential, porting should even be easier (to Linux
> vs to NT; never tried so you can flame me for that statement - there are
> so many similar bad ones in the news from people who write about things
> they never tried).
> A Primergy 760-server with 4 PPro's and 4GB RAM
> (
http://www.sni.de/servers/primergy/prim_us/pr760_us.htm
Do you have Linux running on this thing?
, hope this
> doesn't make my article commercial spam, that's the one I happen to
> know; just as an example for those who only know home-PC's that there's
> more/bigger Intel stuff out there) should be enough for most uses.
>
> A note: the From:-address doesn't mean I know much or anything about
> Oracle's strategy here,
> --> I'm just a guest here @Oracle HQ <--
> (for 6 months) doing a 64bit port of Oracle 8 for our Reliant-Unix
> (SysV, Siemens Nixdorf). Who would come here for doing this on Linux?
How about you? Just bring in your laptop with Linux and start porting!
Or would that fact make your boss and Oracle nasty?
You know whats Todays strategy for portings looks like? First have it
run
on Linux, and that result should then be ported to your SysV reliant
UNIX.
> All the vendors who want to have Oracle available on their platform have
> a couple of people here to help with porting and to provide a link back
> to the people who code the OS. There are so many issues that can best be
> solved by the platform's vendor, e.g. there are changes made to the OS
> (rather than Oracle) in order to support certain features better (e.g.
> make some things faster for the particular Oracle-implementation -
> easier than rewriting Oracle, sometimes, also see below).
>
> Oracle is not so high level like, let's say, a text editor, it relies
> and uses low level functionality a lot. When doing a port, sometimes you
> adjust the OS's behaviour or functionality in order to support an Oracle
> function better, not Oracle itself (often because they did it right or
> because all others already have that OS-functionality, so one can't
> afford not to do it). For this it's good to have an organization to talk
> to, so I'd say if Caldera or someone else wants to be the partner for
> Oracle, they probably have a much better chance than the anonymous and
> fuzzy crowd of Linux users out there (include me there). Either Oracle
> can be convinced that it pays to do the port or somebody else takes the
> risk and does it, here at Oracle and together with Oracle. I don't see
> why _Oracle_ should take the risk.
Why is there a risk at stake when doing a Linux port? Just do it, and
if you like the result release it.
> ***********************************************************************
> Michael Hasenstein; Siemens Nixdorf (Paderborn, Germany)
> currently @Oracle, Redwood Shores, California
> homepage http://www.csn.tu-chemnitz.de/home/mha/
> Linux NAT http://www.csn.tu-chemnitz.de/HyperNews/get/linux-ip-nat.html
> private EMail: M.Hasenstein(at)remove-this rocketmail dot com
> ***********************************************************************
--
++---------------------------++----------------------------------------++
|| R.M. Stockmann || InfoMagic Nederland VOF
||
|| st...@infomagic.nl || Unix administration & support
||
|| http://www.infomagic.nl || The Netherlands ||
++---------------------------++----------------------------------------++
250-Linux: A copylefted Unix-like operating system for 80[3456]86,
250- DEC Alpha, Sun SPARC, Sun UltraSPARC, Motorola 68k,
250- PowerPC/PowerMac, ARM, Mips R[3,4]x00, Fujitsu AP/1000+
250- and more to come.
Stephen Harris
Kristian Koehntopp (kr...@koehntopp.de) wrote:
: Two main problems with Linux remain, though. These are a) a very
: heterogenous installed base (different distributions and
: different releases of libraries) and b) a general unwillingness
Not a problem - SCO Oracle for OpenServer 5 is statically linked at install
time by the install routines. Relinking the Oracle libraries is common.
By performing this relinking, we eliminate 99% of library version worries.
: to work with no-source-available type of packages.
Err.. IMHO you are confusing two distinct groups of people here (or two
distinct uses of Linux). Yes, there are a number of people who love to
have the source code (yes, I'm one of them, and it's come in handy in the
past), but there is also a large market out there for _solutions_ who don't
care about the source code (remember - we are talking about comparing like
with like; if Oracle for SCO is acceptable as a solution, then so should
Oracle for Linux). I'm also one of these people! Solutions for problems.
I recognise there are some areas where I just _can't_ have the source code.
: better"). But still a great deal of support is installation
: support - shrink-wrapped installation as on other Unix platforms
: or under Windows does not work as good.
Blugh? Have you installed Unix Oracle on any platform? The install
routines (orainst, and root.sh) are broken so badly... (sorry, quick
divergence into a minor rant about broken Oracle install routines)
Oracle 7.1.6 for Solaris: won't find a valid "oracle" user. Why? Well,
it looks in the password file, then looks in the NIS tables but then
totally ignores NIS+. This also _seemed_ to happen with 7.3.3 and 8.0.3
but I'd dropped NIS+ by that stage. Don't do anything silly like putting
a / in the ORACLE_SID when prompted by the install screens. It accepts
it and then blows up 50% of the way through. Hmm, 7.3.2 put the biggest
load of rubbish into /etc/rc2.d that I've seen (there was no way the machine
could boot afterwards!).
Oracle 8.0.3 for HPUX: doesn't link correctly at install time and results
in binaries linked against development versions of 8.0.2 libraries in
a wierd path (something like /mnt/cdrom/8.0.2/lib/...) I didn't have time
or inclination to try and break the rest of the install routines (HPUX is
fragile enough...!)
Where is oratab today? /etc/oratab? /var/opt/oratab?
/var/opt/oracle/oratab? Same versions on different platforms, different
versions same platforms... listener.ora files also move around, and come
out of the install routine not working properly...
Don't even _LOOK_ at their Web Application Server stuff. Beating your
head against the wall is much less painfull - and ultimately just as
productive.
Oracle for SCO Openserver is the cleanest install I've seen, and that still
has problems. The oracle personal server for NT didn't do too bad in
comparison! Until I came to the ODBC crud :-(
Don't make me laugh with talks of shrink-wrapped Oracle installations. Oh
yeah... Solaris and HPUX are (meant to be!) their largest server platforms.
Maybe one day they'll have an install routine that works!
: certain hostility shown towards the Stardiv developers when they
: discussed the problem in de.comp.os.linux.misc. True, the group
: of people telling them to go away or release source has been
: small, but overly noisy. But this is certainly not the way to
: encourage porting to Linux.
Again, we are talking different markets. I wouldn't expect Oracle to release
a Unix product targetted at dumb-user without a clue. Such a dumb user
couldn't afford the license fees anyway :-)
$$$$$$$$ That's what Oracle are looking for. Where I work, we are one of
the heaviest (non banking) Oracle users in the country. We don't have
big databases - we just do very clever and wierd things with it :-)
We work the Oracle support desk *hard* and have (we've been told)
approximately 3 times more calls logged to our name than any other
organisation. And most of the calls escalate to America because they are
clever problems we've found. We've even had the code writers talking to us
trying to track down the problems. _That's_ hostile in a commercial
environment. But we pay our $$$ license fees and they have to grin and bear
it.
James Youngman
>>>>> "Kristian" == Kristian Koehntopp <kr...@koehntopp.de> writes:
Kristian> That's nothing. ;) The entire Deja News search engine runs
Kristian> on Linux SMP systems (>50 dual PPro currently and moving
Kristian> to dual Pentium IIs). These machines handle over 3.5
Kristian> Million hits per day, with peak load of about 50 page
Kristian> views per second. They have already ditched one operating
Kristian> system for instability and if Linux SMP wasn't reasonably
Kristian> stable, they would have ditched it long ago. The most
Kristian> common cause for a machine failure is power outages. We
Kristian> have seen some instability in 2.0.31, but for our
Kristian> hardware, 2.0.32 is pretty solid.
Which OS got ditched? Did anyone ever write a case study up?
Joel Garry
In article <EM3rH...@spuddy.mew.co.uk>,
Stephen Harris <sw...@spuddy.mew.co.uk> wrote:
>
>Blugh? Have you installed Unix Oracle on any platform? The install
>routines (orainst, and root.sh) are broken so badly... (sorry, quick
>divergence into a minor rant about broken Oracle install routines)
Yeah, I've ranted for years, why can't Oracle understand it ain't good business
to piss off people right after you've taken their check? I guess a few years
of raking in the bucks makes you think you are doing something right.
>
>Again, we are talking different markets. I wouldn't expect Oracle to release
>a Unix product targetted at dumb-user without a clue. Such a dumb user
>couldn't afford the license fees anyway :-)
A few years ago, there was a conference in La Costa looking at the future
of computing or some such. Phillipe Kahn pointed out there is a strong
economic pressure to push pricing down to about $100/seat. At first I
laughed, since I was at the time supporting sites where manufacturers had
quite willingly plunked down hundreds of thousands for Alphas, HP's and
AIX's, in order to get an appropriate bundle of applications. Then I thought
about it for a while...
>
>$$$$$$$$ That's what Oracle are looking for. Where I work, we are one of
Not a secret! :-) However, after I did the above thinking, I posted we should
watch out for a siege mentality, as stockholder pressure to perform becomes
increasingly difficult. Now I think we are seeing it. If Larry cracks,
we might see some really bizarre organizational behavior. If he slides
through somehow again, maybe this will be the wakeup call to admit that there
will be some hard times ahead, and maybe Linux servers aren't such a bad
idea. We can but hope.
>the heaviest (non banking) Oracle users in the country. We don't have
>big databases - we just do very clever and wierd things with it :-)
>We work the Oracle support desk *hard* and have (we've been told)
Kewl!
>approximately 3 times more calls logged to our name than any other
>organisation. And most of the calls escalate to America because they are
>clever problems we've found. We've even had the code writers talking to us
>trying to track down the problems. _That's_ hostile in a commercial
>environment. But we pay our $$$ license fees and they have to grin and bear
>it.
>--
> Stephen Harris
> sw...@spuddy.mew.co.uk http://www.spuddy.org/
>
> The truth is the truth, and opinion just opinion. But what is what?
> * Meeeeow ! Call Spud the Cat on > 01268 515441 < for free Usenet access *
--
These opinions are my own and not necessarily those of Information Quest
jga...@eiq.com http://www.informationquest.com
http://ourworld.compuserve.com/homepages/joel_garry
"See your DBA?" I AM the @#%*& DBA!