On Mon, 25 Jun 2012 12:13:23 +0100, Java Jive <ja...@evij.com.invalid>
Yes, I see your point. It's been so long since this was more than a
minor issue that I hadn't realised it might not be so minor for
everone.
The point I was making was that splitting the OS and apps off into
different partitions (aside from those "Core Apps" that the OS
pre-installs such as wmp, IE and OE and such) means you keep the
compressed OS partition image quite small, a partition backup that is
the minimum required to successfully recover from almost any disaster,
even one caused by a dead disk drive.
You can make backups of the apps partition on a less frequent basis
if you don't mind having to re-install / reconfigure the odd one or
two "New apps" that got added / updated since the last drive D backup.
It's a lot less likely that the sort of disaster you're protecting
against on the boot drive will also apply to the apps drive. It's
really only the OS partition that benefitst from image backups, the
remaining disk volumes can be backed up using more conventional data
backups including the space saving incremental backup regime.
It's worth remembering that any backup scheme is simply to protect
against "The Unthinkable" (and, hopefully, extremely rare) event of
data loss / corruption whether due to disk failure or any other number
of causes.
I've chosen to place the OS and apps onto seperate partitions to
allow me to optimise the benefits of keeping both backed up on their
own seperate backup schedules. Any complications in "Tidying Up" post
restore onto a replacement disk drive will be a relatively minor
'inconvenience' AFAIAC.
If you're in the habit of making frequent changes of installed
software, the "Keep It Simple Stupid" OS and apps on a single disk
volume strategy will probably be a better trade off against the space
and time saving strategy of the seperate OS and apps partitions where
you can afford to make more frequent OS partition backups with
relatively infrequent apps partition backups.
At the end of the day, it's down to personal preferences and, if
you're setting up a system on behalf of a less skilled computer user,
what you think they're most likely to be comfortable with.
>
>> >One more thing. If one has a disk larger than the 128GB limit, it's
>> >worth installing the XP Recovery Console instead of the W2k one, as
>> >that will recognise partitions that go beyond the 128GB limit. WITH
>> >CARE, once the system build is complete and all updates have been
>> >installed, this can be done as follows:
>> >
>> >1) Look in the root of C: and backup NTLDR to, say, NTLDR.W2K
>> >2) Insert an XP Pro installation CD and install the RC by going into
>> >a DOS console and typing ...
>> >
>> ><CD Drive>\I386\WINNT32.EXE /console
>> >
>> >... answering any confirmations appropriately. You can allow it to
>> >search for updated files, though it actually doesn't seem to make much
>> >difference.
>> >3) IMPORTANT - When installation is complete, before rebooting
>> >rename NTLDR to, say, NTLDR.XP, and replace it with a copy of the file
>> >backed up in 1.
>>
>> That bit is redundent. I'm actually using the winXP versions of that
>> file and the
NTDETECT.COM as a means to shave about 4 seconds off the
>> boot time.
>
>IME, Windows 2000 hangs if you don't do this. What versions of NTLDR
>and
NTDETECT.COM are you actually using? Can you give us the version
>numbers and the date/time/size of each?
Unfortunately, there's no version info for these files, just creation
and modified date/time stamps. These I can give you.
Name Size (bytes) Created Modified
NTDETECT.COM 47,564 19/06/2003 12:05 13/04/2008 23:13
NTLDR 250,048 07/04/2012 12:28 14/04/2008 01:01
NTDETECT.COM.2k 34,724 19/06/2003 12:05 19/06/2003 12:05
NTLDR.2k 214,432 19/06/2003 12:05 19/06/2003 12:05
I did, as you can see, take the precaution of backing up the originals
in case this "Speed up fix" was a load of balony (or even some sort of
nasty joke) but the 'fix' does seem to 'work as advertised'.
The speed up was only a mere 3 to 4 seconds at best and clearly not a
solution to eliminating the 40 or 50 seconds scuzzy delay I was trying
to fix. However, even a mere 3 or 4 seconds is an improvement of sorts
and I've not seen any other consequences over the past two months or
so that I've been trialling this 'fix'.
>
>> Unfortunately the "Scuzziness" of SATA drastically
>> increases the boot time in win2k. In my case, I estimate that it must
>> be adding a good (bad, really!) 40 or 50 seconds making for a total
>> boot time from power on of some 90 seconds in spite of the SSD.
>
>Well, that's not my experience. When I changed my lead machine to
>SCSI 160Mb/s, the boot time reduced noticeably, when I replaced the
>SCSI with SATA 150Mb/s there was no appreciable change.
My first experience with protracted boot times with SCSI devices
harks back to the heady days of a P2/350 on a BX440 MoBo with an
adaptec HBA serving a quad speed CD writer and a flatbed scanner.
The boot process was, afaicr, taking about 3 minutes or so until I
sorted out the SCSI adapter settings which got it down to less than a
minute. Unfortunately, there are no such settings available with the
built in SATA controller to allow me to fine tune it to prevent win2k
seemingly going through a rather pointless 'discovery' exercise during
the time it displays its start up logo screen.
>
>> Despite this extra 40 odd seconds required to boot, I still prefer
>> win2k's better responsiveness and lower maintainance requirements to
>> the rather dubious 'delights' of winXP.
>
>Yes, I think there's more bloatware to disable in XP.
>
>> As far as installing a recovery console, I find it all rather
>> unnecessary.
>
>My experience is the opposite.
>
>> In fact, the relatively frequent (around once every
>> couple of months or so) Licence violation errors that are caused by
>> using an SSD make such a recovery console redundent since fixing FS
>> errors on the SSD have had absolutely no effect whatsover on this
>> particular error which can only be fixed by restoring from an image
>> backup file thus neatly demonstrating that an image restore will trump
>> winXP's rather fragile system restore feature every single time.
>
>That seems to be an argument against using SSD drives rather than an
>argument against installing the recovery console.
Well, it's an argument against _this_ particular SSD (a "30GB"
Kingston SSD Now V series drive). I don't know whether it's an
argument against _all_ SSDs in general (at least not yet).
The point here was that it was a failure mode I couldn't have
predicted (at least not without a damn good reason to research very
deeply into the many issues new to this technology) yet the partition
image backup strategy proved itself equal to the job regardless of
whatever form the disaster took.
Prior to using an SSD, I can't remember any event where my only
remedy was to restore from an image backup. I'm sure there must have
been two or three times over the decade prior to my fitting that SSD
where I was forced to restore from a disk volume image but I can't
recall any one particular event, just a vague notion that I must have
had no choice other than use a backup image.
Of course, I've restored from image backups where I've made the
mistake of installing questionable software as a guaranteed means of
rolling the system back to "Happier Times".
SSDs have serious issues over handling a reset during impending loss
of power regardless of whether it's a power outage or a system
shutdown, issues which simply do not exist for "Spinning Rust"[1]
drives.
A spinning HDD only has to, at most, terminate a sector write in
progress gracefully to avoid spurious 'bad sectors' and co-lateral
data corruption, a matter of microseconds, whereas an SSD has to
perform a cache flush to an erase block or three (maybe more) which
takes tens to hundreds of milliseconds. This requires a level of
reserve power 4 or 5 orders of magnitude longer in duration over and
above the HDD requirement which can be very comfortably exceeded by
the voltage supply hold up time from the main PSU during the time the
PG line becomes negated (triggering a system reset condition) and the
voltage rails eventually dropping below the minimum level for
guaranteed operation of the logic gates used in the HDD's controller.
An SSD aught to be able to independently provide its own reserve of
power to ride out such a shutdown event sufficiently to completely
avoid the risk of data corruption, especially since such data
corruption can effect totally "innocent data content", not just that
of the file caught out in 'mid flight'.
That 30GB Kindston SSD either relies upon the main PSU hold up
reserve time or else has insufficient built in reserve. Either way,
the reult of this lack of reserve has proven to be disasterous on far
too many occasions for my liking over the past couple of years that
I've been 'trialling' the SSD. Admittedly, the situation did improve
somewhat after I upgraded the PSU, leading me to believe this
particular model of SSD was more or less totally reliant on the main
PSU reserve rather than on a self contained one.
[1] an expression I've seen used by a certain poster using a two
decades out of date technology reference.