SBS Mirrored Servers

[David Taylor]

We have a client that is asking whether it is possible to used a mirrored
disks of the SBS server in another machine (all be it identical) if they
have a hardware failure?

Thanks
Dave

[Steve Foster [SBS MVP]]

If you had two identical servers, and wanted to swap hard drives
containing a working SBS2000 from one to the other, that would work.

Mind you, they would need to be exactly identical down to the BIOS
version and all main motherboard components, not simply two servers of
the same make/model (compaq were good at doing this - two identical
make/model servers, even bought together, could have different
innards!).

It's a pricey way to limit downtime... Are they sure it's necessary?

--
Steve Foster [SBS MVP]
---------------------------------------
MVPs do not work for Microsoft. Please reply only to the newsgroups.

---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.404 / Virus Database: 228 - Release Date: 15/10/2002

[Gary Smith]

It worked for me on SBS4.5 and I have bought 2 identical (I hope) Dell
servers for SBS2k with this backup scenario in mind. In the meantime I use
the second server as a workstation with a smaller harddisk. I haven't had to
try the brain transplant on SBS2k yet.
"Steve Foster [SBS MVP]" <steve....@picamar.co.uk> wrote in message
news:OeWh4iQgCHA.3596@tkmsftngp08...

[David Taylor]

Steve, Gary,

Thanks for the post.

Just to clarify this more. The user has an SQL database that is "mission
critical" to the business. As normal they do not want to pay thousands of
pounds for technology. But at the end of the day its that old friend,
"mission critical" data. How much are they going to lose if they are down
for a period of time.

They currently have a server from Dell which they had the with warranty but
Dell could not get a motherboard for a couple of days. To the customer this
was no good. They had to be up 100% of the working day.

From what I have been think about in the last few days the only options
available are below. Please bear in mind that I have no experience with
how SQL is ran / restored etc, if some has a better method then please
shout, but not too loud :)

1 - Get 2 identical machine (Dell's). Ghost the server, reload this image
on the spare machine and then restore from tape the SQL database. Would
this work...? I have no experience with how SQL is ran / restored etc...
What about licensing of SBS....?

2 - The customer buys another copy of Win2K SRV and SQL with CALS that he
can run on the separate server and then restore from tape as and when
required. To me this second option is rather defeating the object of why
they bought SBS 2K. It has all the product that they want for the right
price. How easy would this be to setup and configure?

Is it possible to cluster an SBS server? What about having 1 machine name
but 2 hardware boxes configured in a RAID 5 sort of solution? Eg If the
first machine fails the 2nd carries on. Both machines are running and
updating at the same time. Anyone had any experience of this before?

If anyone has another ideas then please speak now, for tomorrow may be too
late.

Thanks
Dave

"Gary Smith" <gary...@cummingsandsmith.com> wrote in message
news:eOUwIXSgCHA.1636@tkmsftngp10...

[Gary Smith]

When you say the SQL is mission critical you have to think about the amount
of time which you can be down for without serious business consequences and
then design the system around that. For example, electrical failures of a
few minutes or less are handles by a battery ups. Hard disk failures can be
handled by mirrored drives or raid 5 and tape backups which probably at
worst puts you down for a couple of hours or often much less. Failures of
other system components such as power supplies, network cards, etc. can be
handled by have spares on the shelf as these are not costly and again you
are down only the time to replace the failed component.

The scenario you described, a failed motherboard, was what drove me to buy a
spare server in the SBS4.5 days. I bought a spare server, same type, same
brand, same time from Dell, but I didn't populate the RAM as fully and I
bought only one hard drive for the spare. I use it as a workstation. I use
mirrored hard drives with tape backup on the main SBS server, I'm not sure
that it will work, but my plan is that if the main components of the working
SBS server fail (motherboard or other critical component) I will simply move
the hard disks and potentially some ram chips from the failed server to the
spare and within a couple hours I am back in business, what I call a brain
transplant. I can then replace the failed parts on a more leisurely pace. I
think my cost for the spare server was about $2000 US and for that I got to
use it as a spare workstation for what is not much more than the cost of a
regular workstation.

Again, think about the amount of down time you can tolerate and then think
through all critical system components in that light. I hope this helps.
"David Taylor" <nosp...@amshire.co.uk> wrote in message
news:uMiiC7YgCHA.2400@tkmsftngp10...

[SuperGumby]

Guys,

an associate does something I consider unusual which may pertain to this
conversation. We perform work for him on occassion.

The first box he had configured which I encountered had no internal
HDD's. SBS running off a pair of software mirrored SCSI's in a box connected
to the external interface of the SCSI controller. My initial reaction
probably shouldn't be decribed in public. Then I thought about it.

If he attaches those two drives to a simalarly deviced box he can handle
a 'catastrophic failure' which takes out anything but the drives. Matter of
fact, he can lose one of the drives.

I'm still not comfortable with the idea but I'm suggesting it may be food
for thought.

[Steve Foster [SBS MVP]]

I have a client for whome the SQL system I designed & maintain is
"mission-critical". In fact, without it his business doesn't function
_at all_.

He had a motherboard failure in his SBS4.5 approximately 2 years ago. He
was down for 1 day. I was there all that day - we got a replacement
motherboard from the manufacturer and collected it ourselves.

Once that was obtained, it took me a couple of hours to get him back
fully operational.

Afterwards, he asked me how to prevent it ever happening again and we
discussed various options. At the end of the day, we decided that the
critical factor is his data. As long as we have access to that, we can
always get him up and running in some fashion within a couple of hours.
If necessary, I can run his systems from one of my servers, translocated
temporarily, or we can zap one of his other PCs (either another server,
or even a workstation). Heck, he has a PC superstore less than 30
minutes drive away.

The key question is what is the business cost of having no server for a
day, half a day, 2 hours, etc. And then how much money are they prepared
to spend to mitigate that.

Version: 6.0.410 / Virus Database: 231 - Release Date: 31/10/2002

[Buddy Greenshield]

For truly mission-critical SQL applications, SBS is the wrong platform as
you need the advanced server and SQL enterprise versions to set up
clustering. You can setup a warm standby SQL server using replication
techniques, but you wouldn't be able to use the SBS license for this, and
then you get into the CALs' issue as well. A standby server is a good
alternative. There is also a unique software solution that provides
automatic failover.. Double-Take, worth checking out.
http://www.sunbeltsoftware.com/product.cfm?id=111

Buddy Greenshield

"Steve Foster [SBS MVP]" <steve....@picamar.co.uk> wrote in message

news:#bpRO7bgCHA.1004@tkmsftngp08...