Disaster recover recommendations

  • Thread starter Thread starter Bob Rogers
  • Start date Start date
B

Bob Rogers

I am a consultant and my clients are all small businesses (1-5
servers, 10-150 workstations) We tend to buy new servers as we need
them, so there is no real consistency between operating system, hard
drive types or size and of course applications installed on them.

What we want is an ability to image these servers (and workstations as
well, but that is less of a problem). We will have a "backup" server
with removable hard drive cages that we will clone the image to. This
backup server is nothing like the original computer. They may be 3
years apart in manufacture. This harddrive cage then will be
squirreled away off site. We will also use incremental (or even new
if we have to) images periodically just to keep current. If the
original server crashes (as in the computer dies, not just the hard
drive has to be replaced). We want to apply the incremental images to
the "backup" system and then replace the dead system.

After that we will buy a new system, and clone to THAT machine and
move that one into production, returning the "backup" system to a
disaster recovery status

1. Am I asking for the impossible here with ANY product?
2. What happens if the original server had mirrored drives and the
"backup" does not?
 
Success will depend on how much effort you are willing to put into that
project.
The biggest chalenge is to have a replacement server/workstation up and
running quickly.
Are you planning to have an idling swap computer on each site?
Assuming that hardware of that swap system will differ from failed system,
are you willing to prepare and test detailed recovery procedure for each
important machine? That is a lot of work (sysprep, QA) not just clone disks.
Having just a bunch of disk images is not good enough.
If the original server has mirrored drives, then it depends on imaging
software and hardware of the replacement system. Usually imaging software
does not care about underlying RAID (unless it is a software or firmware
RAID).
Your strategy would be quite efficient for disk failures only. But if
unique/old RAID controller dies and you do not have a spare one, I wish you
luck.

For small businesses I would suggest to focus on protecting data and being
able to install critical applications quickly from scratch, using new
hardware, of course. Minimize system failures by constant monitoring,
replace older systems regardles from their health record.
 
Previously Bob Rogers said:
I am a consultant and my clients are all small businesses (1-5
servers, 10-150 workstations) We tend to buy new servers as we need
them, so there is no real consistency between operating system, hard
drive types or size and of course applications installed on them.
Click to expand...
What we want is an ability to image these servers (and workstations as
well, but that is less of a problem). We will have a "backup" server
with removable hard drive cages that we will clone the image to. This
backup server is nothing like the original computer. They may be 3
years apart in manufacture. This harddrive cage then will be
squirreled away off site. We will also use incremental (or even new
if we have to) images periodically just to keep current. If the
original server crashes (as in the computer dies, not just the hard
drive has to be replaced). We want to apply the incremental images to
the "backup" system and then replace the dead system.
Click to expand...
After that we will buy a new system, and clone to THAT machine and
move that one into production, returning the "backup" system to a
disaster recovery status
Click to expand...
1. Am I asking for the impossible here with ANY product?
2. What happens if the original server had mirrored drives and the
"backup" does not?
Click to expand...

Just one comment: With Windows I have made very bad experiences
when any part of the hardware changed. From minimal nuisance to
complete failure to boot.

With Linux you can usually rip out the disk, put it into another
machine and it will work fine or only need minor adjustments.

I have done this myself several times successfully with Linux. I have
had to struggle with Windows for hours...days on basically every
mainboard change, sometimes even on a CPU change.

For incremental backups on Linux/any unix you can just use "tar" with
the date option. On Unixes other than Linux you can also use "dump"
for incremental backups. You cannot however restore the incremental
data to the system partition in a live system. It is a good idea to
have a recovery system (extra system partition that is booted into for
recovery) on your disk.

One problematoc point is if you have any specialized hardware
like RAID controllers. You often need the same controller
to be able to use the disks then.

Of course that is for servers. For workstations you get the
whole "multimedia" questions (video, sound, Mice, etc.) and
there Linux can be as problematic as Windows.

My guess is that you will actually need to look at each OS and
installation separately and figure something out that works.

And, yes, it is definitely a good idea to be prepared for
loss of hardware without the possibility to get the same
hardware again.

Arno
 
Bob Rogers said:
I am a consultant and my clients are all small businesses (1-5
servers, 10-150 workstations) We tend to buy new servers as we need
them, so there is no real consistency between operating system, hard
drive types or size and of course applications installed on them.

What we want is an ability to image these servers (and workstations as
well, but that is less of a problem). We will have a "backup" server
with removable hard drive cages that we will clone the image to. This
backup server is nothing like the original computer. They may be 3
years apart in manufacture. This harddrive cage then will be
squirreled away off site. We will also use incremental (or even new
if we have to) images periodically just to keep current. If the
original server crashes (as in the computer dies, not just the hard
drive has to be replaced). We want to apply the incremental images to
the "backup" system and then replace the dead system.

After that we will buy a new system, and clone to THAT machine and
move that one into production, returning the "backup" system to a
disaster recovery status

1. Am I asking for the impossible here with ANY product?
www.acronis.com

2. What happens if the original server had mirrored drives and the
"backup" does not?
Click to expand...

RAID 1 works on just one drive so that's fine.
 
One problematoc point is if you have any specialized hardware
like RAID controllers. You often need the same controller
to be able to use the disks then.
Click to expand...

Not true so much for RAID 1 whether HW or SW/firmware.
And, yes, it is definitely a good idea to be prepared for
loss of hardware without the possibility to get the same
hardware again.
Click to expand...

The number one issue that image backup protects against is HD failure,
physical HW failure or failure of the contents(corruption). In windows an
image restore to the same or different but similar HD virtually always works
with no hassles. Beyond that a good image backup allows individual files
access/recovery if the restore on new HW really doesn't want to work. In XP
and W2K3 one can almost always make it work on new HW BUT sometimes that
process is a greater hassle than OS & app re-install plus individual file
recovery.

I keep a second PC HW indentical to the server PC for modest configurations.
That 2nd PC is used as a workstation. Server box breaks and just swap HDs
and some cables and one is very quickly good to go.
 
Ron said:
The number one issue that image backup protects against is HD failure
Click to expand...

The number one reason for using IMAGE backups over any other backup
solution is to save time with OS & app reinstall & configuration.

An image restore can easily save a whole day's restore time.

If (like the OP) you are restoring to vastly different hardware (3 year
difference was mentioned) then the OS is unlikely to handle it,
so you'd be faced with a longer restore time (and a less stabel server)
than an os reinstall & file based restore.
 
The number one reason for using IMAGE backups over any other backup
solution is to save time with OS & app reinstall & configuration.
Click to expand...
An image restore can easily save a whole day's restore time.
Click to expand...
If (like the OP) you are restoring to vastly different hardware (3 year
difference was mentioned) then the OS is unlikely to handle it,
so you'd be faced with a longer restore time (and a less stabel server)
than an os reinstall & file based restore.
Click to expand...

As I said, that vastly depends on the OS. I have succeeded in booting
a PC with a notebook installation (Linux) without any changes. I have
had an unbootable system (crash on boot) with Windows after just
changing the CPU. I have no idea how other OSes behavie in
this regards.

Arno
 
Not true so much for RAID 1 whether HW or SW/firmware.
Click to expand...

Unfortunately even true for RAID1 sometimes. The contoller
has to identify which disks/partitions belong to a set (if
there is more than one possibility). This may require
a descriptor block and there is the problem, since that
can be in different locations. Mist problematic: Descriptor
at the beginning and controller translates sector numbers
to hide it. Least problematic (no problem usually): At the
end and controller makes disk a little smaller.
The number one issue that image backup protects against is HD
failure, physical HW failure or failure of the contents(corruption).
In windows an image restore to the same or different but similar HD
virtually always works with no hassles.
Click to expand...

That is my experience also. Image backups of Windows can
be copied to other harddisks. You may need a rescue-floppy
to make them bootable, but that is it.
Beyond that a good image backup allows individual files
access/recovery if the restore on new HW really doesn't want to
work. In XP and W2K3 one can almost always make it work on new HW
BUT sometimes that process is a greater hassle than OS & app
re-install plus individual file recovery.
Click to expand...

Yes, my experience also. You can get it to work, but itmay take a
day or so and it is painful. At least for somebody who is not
a Windows expert (like me).
I keep a second PC HW indentical to the server PC for modest
configurations. That 2nd PC is used as a workstation. Server box
breaks and just swap HDs and some cables and one is very quickly
good to go.
Click to expand...

If you can do this, it is possibly the best solution for Windows.
Getting old hardware can ge difficult theoug, to come back to
the OP's situation.

Arno
 
I'm trying to find out if Universal Imaging Utility would take care of
the driver problem when cloning a workstation.

Do you know anyone who has used it or how it works?
 
:arrow: I have used the Universal Imaging Utility and it works very
well for workstation cloning. You should download the free trial
from www.binaryresearch.net and check it out for yourself. It saved
me a ton of time and I only have to maintain one Ghost image now.
 
Mark Jasterwrote:
I'm trying to find out if Universal Imaging Utility would take care
of
the driver problem when cloning a workstation.

Of course that is for servers. For workstations you get the
whole "multimedia" questions (video, sound, Mice, etc.) and
there Linux can be as problematic as Windows.


I use the Universal Imaging Utility all the time. Of course, I
Click to expand...
helped in its development and I support it. It should work with most
hardware; however, if you have something with brand spanking new
hardware that we haven't been able to get the new drivers for, you
may need to download an update from us. Feel free to contact me:
www.bigbangtraining.com

Do you know anyone who has used it or how it works?[/quote:063e02d22b]
 
Back
Top