looking for opinions on hotswap SATA drives as a viable backup solution

[Jason]

I'm in the process of putting together a new backup solution for one
of our file servers and I wanted to throw my idea out there and see
what sort of flaws there are in the plan, that I can't see myself.

We've got a server that has outgrown its current tape backup (40/80
dlt). The server will eventually pass 100gb of data. If I want to stay
with tapes, I have to move up to a 100/200 LTO option which puts the
drive cost alone at about $3000+, plus tapes. While this is definitely
a viable option, it's still somewhat limited. The data we have on this
server isn't going to compress much more than it already is, so
basically when we hit 100gb, we have to shell out for another new tape
system.

My plan was to setup hotswap SATA drives and use backup to disk
folders in backupexec. More specifically I would probably go with a
good SATA raid controller (raidcore or 3ware), a single SATA removable
enclosure (+ trays for each drive) and 10 160gb SATA drives. This
would give me a ten day rotation of backups with plenty of future
space as our needs grow. It also allows for easier upgades in the
future, once we surpass the 160gb barrier the current drives would
have. All this for under $2000.


I am well aware of the potential hazards of using HD's as a backup
media as far as shock is concerned. If someone accidently drops one of
the drives, we could very well have lost that days backup. Beyond
this, are there any other potential problems anyone can point out in
this plan?

Also, if anyone has a better solution, please feel free to share it.
From where I sit, tape backup is a dead horse. Average storage needs
in the industry has far outpaced what tape backup options can deliver
on a cost/size basis. I'm trying to be as progressive as possible
without making more problems for myself down the road. Thanks!

 

[Mr. Grinch]

I think this is workable and have thought about doing the same thing. In
fact I've often used large hard drives as temporary backups during
migrations of data. Together with Robocopy, it's been a very workable
solution. Prior to migration day, I let Robocopy run in a continuous
loop to sync the data from the source to the backup drive. Then on the
day of migration, the final sync takes very little time.

Point being, you don't have to do your backups all at once, you could
sync data all day long then run a final sync at night which will be much
quicker.

You might be able to also use Shadow Copy technology from Server 2003 to
do some neat things but personally I know I can rely on Robocopy.

My only complaint with your backup solution is the schedule. I would want
one drive for every day of the week, 4 drives to set aside for each
weekly, and 12 drives to set aside for each monthly.

As with any backup solution, it's only as good as it's ability to
restore. This is where you're going to have to test it out to see if
your retention schedule and restore procedures really work in getting the
right data from a given date and how long it takes.

There is another option to you to consider. I would encourage you to
look at the large scale corporate near-line solutions. One of the
features they offer is to backup to the near-line storage (slow big hard
disk arrays) then from there, they back up to tape. As a result, they
don't have to worry about open file issues, and can take much longer for
the tape backup because it can run all day long if required. Then you'd
have two backups. Something worth considering depending how valuble the
data is or how expensive it is to replace it.


(e-mail address removed) (Jason) wrote in @posting.google.com:

 

[Tom]

The hot swappable feature requires a rail insertion system to a backplane,
or there is a huge risk of damage to the drive.
This was intended for servers, etc., not the home pc.