On Thu, 06 May 2010 10:06:58 -0400
jasonswett <[email protected]> wrote:

> 
> I read through a good portion of W.Curtis Preston's book but I'm
> afraid I'm too much of a beginner to know where to start with my
> solution. I think I want to use Amanda but everything I've read on
> the Amanda site is too narrow and specific.
> 
> Here's what I have:
> 
> I have two Linux boxes. One is a development server and one is a
> production server. Each is running a database instance that I want to
> back up.

If the devel DB server is a copy of the production one, you don't need
to back that up. If you have to restore that machine, rebuild as
needed, then copy the production DB over as usual. On the other
tentacle, storage media is cheap; it may be easier to back them both
up.

We are rather floundering around in the dark until we know how much
data you have to back up, and how often it should be backed up.

> 
> I also have two Macs and one PC that are just workstations.
> 
> I've looked into Amanda a little bit but just came out confused. I
> don't have any storage media but I'm willing to buy some.

How much data do you have to back up? The production DB, related
software, the devel environment, and the two Macs and the PC. That
tells you how much data you have to back up. Divide those up into DLEs
(I forget what DLE stands for, but it's the unit of backup).

Now, how often do you want to start each DLE over again with a total
backup? Add in a generous fudge factor for how much changed data you
expect between totals. How many complete such cycles do you want on
hand? Multiplying those gives you a rough estimate of the capacity of
backup storage you'll need. Again, add a generous fudge factor.

Divide the total amount of data to be backed up per cycle by the number
of backups per cycle, and you have a rough estimate of the data to be
backed up each run. Divide that by the data rate of your media or
network, whichever is slower, and throw in a fudge factor, and you have
how long each run should take.

If you can get all that onto one hard drive, I'd go with a separate hard
drive and vtapes for storage media. Even if you need two or three hard
drives, they might be cheaper and more reliable than tapes. I guess
that you could have a configuration for each hard drive.

You may also want a separate configuration for the production DB,
depending on how much data you are willing to lose. That way you can
back it up more often than you back up everything else. E.g. you may
want to back everything else up daily on weekdays. But if your database
is the lifeblood of your business, and it's an ecommerce business, you
may want to back that up twice a day every day, 2x7.

Amanda assembles its backups on a holding disk; that should not be on
the storage drives to improve performance.

I have a small installation, and I use external USB hard drives, but you
may need internal drives for the speed, depending on how much data you
have to back up each run.

I would do this incrementally. I would get the two Linux boxes backed
up, except for the two DBs (they have their own issues). Get those
working and be sure you can restore from the backups.

That's the approach I'd take; I'm sure other folks will weigh in with
their ideas. But get us some numbers and we can better help you.

-- 

Charles Curley                  /"\    ASCII Ribbon Campaign
Looking for fine software       \ /    Respect for open standards
and/or writing?                  X     No HTML/RTF in email
http://www.charlescurley.com    / \    No M$ Word docs in email

Key fingerprint = CE5C 6645 A45A 64E4 94C0  809C FFF6 4C48 4ECD DFDB

Reply via email to