[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

my backup and redundancy policy



PureBytes Links

Trading Reference Links

I have Ghost, Casper, and a few other imagers, and I've used them all in
real-world situations. The eventual conclusion and redundancy design I've
come to is this.

Any PC that's critical, implement RAID1.

Automate incremental backups (each night) to a server. For (very) critical
data, do real-time backup from PC to server.

Run DoubleTake on the server(s) for real-time redundancy. If a server fails
another server immediately takes over. Also implement DoubleTake on a remote
server. If a local disaster occurs, you can remain in business from any
other location.

Tools like Ghost and Casper fail miserably when you're under pressure.
Casper is useless if you have to use incompatible or non-identical hardware
to restore, which is usually the case after a few years go by. For example,
you can't easily move a partition to another manufacturer's disk with quite
different (albeit generic) specs. You'll find it's a real pain and it's very
easy to loose an MBR or the contents of a partition becomes unrecognizable
with a different version of an operating system. Ghost probably works well
if you have the intellect of a rocket scientist and you don't make a single
mistake--otherwise it's a pig of a thing.

As I mentioned one time before, Partition Commander is my favorite. Very
flexible and easy to use. To make images safely imo, you should always boot
a PC from a diskette or a bootable CD that you've made. Chain an additional
disk to the cable and remove all jumpers from it so that it's a slave.
Leaving a jumper on and relying on cable-select mode can be problematic.
Forget about imaging to a target over a LAN with the kinds of tools
mentioned. They really don't do that with any in-process recovery mechanism.
If you have a bad packet collision, you'll never know about until you go to
recover, and then its too late.

A gotcha to look out for. Always have an identical spare disk on the shelf
for a hot swap if a RAID fails. NOT a similar disk as RAIDs won't always
rebuild on different hardware. If that happens, then you'll have to "undo"
the RAID, image the good disk to a new disk, plug in the new disk and a
second (blank) identical disk, then rebuild the RAID. It's not that hard
after you've done it 2-3 times :).

In summary, I believe RAID is the most appropriate storage solution, and
backup to another computer according to criticality, either incremental or
real-time.

Colin West