Disaster Recovery Scenarios?

Howdy

Wonder if anyone has any suggestions for a disaster
recover scenario for a customer of mine.

They are running SBS 2000 on a server with ide raid 5, as
well as a DDS4 tape drive. They run regular backups but
are concerned about a critical failure of the server that
might have them down for a period of time.

They would like to have a PC that in the event of their
server failing could have SBS restored to it in a short
space of time as an interim measure until the server is
back on line. Any suggestions of the best way to do
this? Some of the scenarios that I would like comments
or experiences people have had are:

1. Backup up SBS regularly over the network to a file
and then restore this backup to the PC in the event of a
failure. Problems I see with this are the short file
name issue and restoring to new hardware.

2. Using some kind of snapshot program (ghost or
similar) regularly to create an image over the network to
restore in the event of failure. Again I think restoring
to new hardware might be the issue.

If anyone has a suggestion of an applications, hardware
or methods to do this I would appreciate it.

Also any obstacles I may face trying each of my scenarios
would be appreciated.

Regards
Ian

Grey
Thu Jul 10 06:28:26 CDT 2003

ghost to same hardware is the only reliable thing that comes to mind

--
Grey
SBS ROCKS MVP


www.smallbizserver.com


"ILR" <iragless@hotmail.com> wrote in message
news:086e01c346ba$e2d706b0$a501280a@phx.gbl...
> Howdy
>
> Wonder if anyone has any suggestions for a disaster
> recover scenario for a customer of mine.
>
> They are running SBS 2000 on a server with ide raid 5, as
> well as a DDS4 tape drive. They run regular backups but
> are concerned about a critical failure of the server that
> might have them down for a period of time.
>
> They would like to have a PC that in the event of their
> server failing could have SBS restored to it in a short
> space of time as an interim measure until the server is
> back on line. Any suggestions of the best way to do
> this? Some of the scenarios that I would like comments
> or experiences people have had are:
>
> 1. Backup up SBS regularly over the network to a file
> and then restore this backup to the PC in the event of a
> failure. Problems I see with this are the short file
> name issue and restoring to new hardware.
>
> 2. Using some kind of snapshot program (ghost or
> similar) regularly to create an image over the network to
> restore in the event of failure. Again I think restoring
> to new hardware might be the issue.
>
> If anyone has a suggestion of an applications, hardware
> or methods to do this I would appreciate it.
>
> Also any obstacles I may face trying each of my scenarios
> would be appreciated.
>
> Regards
> Ian
>
>
>
>
>


---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.493 / Virus Database: 292 - Release Date: 6/25/2003

Mark
Thu Jul 10 07:06:34 CDT 2003

ghost to USB hard drive. This can be scripted and run on scheduled
intervals. DDS is not an ideal backup media, DLT is much more
stable/reliable.

--
Sincerely,
Mark Mancini, CCA, CCNA, Master CIW&CI, CNE 4&5, MCSE+I 4&2000
www.MCSE2000.com
www.AppLauncher.com



"ILR" <iragless@hotmail.com> wrote in message
news:086e01c346ba$e2d706b0$a501280a@phx.gbl...
> Howdy
>
> Wonder if anyone has any suggestions for a disaster
> recover scenario for a customer of mine.
>
> They are running SBS 2000 on a server with ide raid 5, as
> well as a DDS4 tape drive. They run regular backups but
> are concerned about a critical failure of the server that
> might have them down for a period of time.
>
> They would like to have a PC that in the event of their
> server failing could have SBS restored to it in a short
> space of time as an interim measure until the server is
> back on line. Any suggestions of the best way to do
> this? Some of the scenarios that I would like comments
> or experiences people have had are:
>
> 1. Backup up SBS regularly over the network to a file
> and then restore this backup to the PC in the event of a
> failure. Problems I see with this are the short file
> name issue and restoring to new hardware.
>
> 2. Using some kind of snapshot program (ghost or
> similar) regularly to create an image over the network to
> restore in the event of failure. Again I think restoring
> to new hardware might be the issue.
>
> If anyone has a suggestion of an applications, hardware
> or methods to do this I would appreciate it.
>
> Also any obstacles I may face trying each of my scenarios
> would be appreciated.
>
> Regards
> Ian
>
>
>
>
>

Jeff
Thu Jul 10 09:08:09 CDT 2003

I think that anyone running an SBS should plan to lose a day of work to
recover a dead condition.

If a down day is too long to be down entirely, then add another server to
the network as another DC, and plan to lose some of the functionality of the
network for a full day. Also plan for you daily operations cost to increase
as a result, at least in some manner. Plan that which ever server hosts the
data, printers, references to UNC pathed server applications, email and web
connections will take out of service those items while that server is down.

I believe that the majority of IT people who purchase additional server
hardware for an organization to have sitting on the side "in case" of a
disaster, and yet don't actually practice recovery every couple of months,
probably are misleading this company into false sense of security. The more
you know about how many ways you can lose functionality, and what it takes
to recovery it, the more you realize that the problem has less to do with
hardware on the side than it does with skill in the IT guy.

There are distinct characteristics of a disaster recovery that have to do
with:

- motherboard HAL similarity/compatibility to the previously backed up
system
- boot controller similarity/compatibility to the previously backed up
system
- recovery of the System State restore, and isolating the hardware issues
that can cause conflicts with the NICs to occur
- recovery of AD involves either a System State restore, or a sync from an
authoritative DC still online
- recovery of Exchange involves maintaining the same AD, or same server
name, domain name, organization name if you rebuild from scratch

In general, any SBS server can be recovered from totally different hardware
if you take even modest care to preinstall drivers for a boot controller you
plan to use in recovery, and ensure that the HAL of a motherboard you will
use is compatible to that of the machine you are replacing. That HAL issue
sounds scary, but there are far fewer HALs out there than most people
realize. Most dual processor motherboards from one manufacturer are probably
HAL compatible to any other one from that maker. To find out, you can
inspect the device manager, or just try booting a drive image on the other
machine.

As such, the most practical viewpoint of a disaster recovery is determining
the fastest way to get a drive image of a previous condition of the original
server, making it operate on the new server, the restoring the System State
to last backup, finally restoring data from last backup. A key point would
be to prepare the stuff you can do in advance. You can test a drive image
boot on different hardware whenever you want before the problem. You can
test a System State restore as well.

Probably the simplest way to prepare a really good disaster recovery
scenario with fast recovery is:

- prepare a drive image on a convenient drive which, honestly, doesn't even
need to be the same controller type as long as the boot controllers and
boot.ini options allow it to boot.
- confirm that you can use a non-critical workstation in your office to boot
this image, or else dedicate a server class replacement system and install
this drive as secondary drive, using a method that allows you to easily
reconfigure it to become the boot drive. (for instance, make the primary
drive removable, or swappable in a bay with this secondary drive). This is
intended to make the imaged drive copy a volume you can access, but not the
one running the computer on a daily basis.
- On a nightly basis, have the normal SBS server perform a System State
backup redirected to a file, and locate that file on the recovery drive in
the recovery computer.
- Ensure that you have the means to easily attach the same tape drive, or a
different compatible one to the recovery computer.

With these preparations, you can probably recover the SBS on different
hardware in the time it takes to reboot this recovery computer, restore the
System State condition last recorded to file on the "backup to file" copy on
that drive. From there, perform a data recovery from tape.

If you want an expedited recovery faster than that, restore each backup
nightly to another drive in this same machine in such a way that you always
have the "previous end of day" 24hrs back stored on this drive. If you make
nightly backups that include a backup of "just files changed today", you
could perform a restore of that backup. You might even consider making
nightly backups to disk of files that change today, plus a backup to disk of
the SBS System State, and on a daily basis you are restoring it to this
server making this machine a constant "one day back" condition.

If you like the idea of the "one day back" restores, you can increase the
frequency to "half day back" and do it twice a day, if the data can be
backed up and restored easily in a 4hr period.