Mark Schoonover wrote:
Les Mikesell wrote:
Mark Schoonover wrote:
Les Mikesell wrote:
Mark Schoonover wrote:
John R Pierce wrote:
Aron.Darling@xxxxxxxxxx wrote:
Loader are totally a love/hate relationship. They do make life a
lot easier as they do the tape movements for you which can be a
tedious thing at times. With a loader or library you can script
the entire operation with tar, MTX and MT and let cron do all the
work for you. Always look for the OEM rather than buying the name
brand equipment, they are most always the same HW and FW with a
different model number in it.
otoh, its hard to beat a 3 year warranty and on location support
from the same vendor as your server hardware, assuming your a
brand name shop in the first place.... hugely reduces finger
pointing when there's a complex issue to resolve. with OEM
hardware bought on the whitebox market, you're often faced with
replace or self-repair option at cost.
Having used a 20 tape library, and suffering through restores with
AIT2 tapes taking 10-12 hours per tape, I gave up on them. I went
with good old rsync, and built up a 4 TB system to handle backups.
Once configured, it's nearly a 100% hands off solution. You can
read about what I've done here:
http://marks-tech-pages.blogspot.com Works great especially for
TBs of data that needs to be backed up every day.
If you want something that stores the backups much more efficiently
(with a price in processing to do it), look at backuppc:
http://backuppc.sourceforge.net/
It compresses everything and hardlinks all duplicates so you can
keep about 10x what you'd expect online, and it has a nice web
interface for browsing the backups and doing restores.
Thanks Lee. I did look at backuppc before, and I didn't want
anything that compresses files, or used a web interface. Using
rsync, it's a matter of scp to restore, and that's it. I do use
hardlinks to duplicate data, so my storage requirements are kept as
small as possible. Just about any CentOS system can be configured to
run backups in the manner I've written up, nothing extra to install
or learn. So, like most things, it boils down to individual needs
and expertise. I know for myself, tape backups just weren't working
well at all.
Backuppc can be configured to not compress, although I think the
filenames are still somewhat mangled and don't have their real
attributes so you can't access them directly. The web interface is
also optional and there are command line tools for everything you
need. It is handy to be able to download a single file or tar/zip
archive directly through a browser, though - and in the latest
version you can edit the configuration through the web interface.
I've thought about adding a web restore so users can restore their own files
to their own home directories. Most of my problems come from my graphics
dept, where it's not uncommon for them to 'accidently' delete 500GB of data.
It would be nice to offload those kinds of requests to the users, but only
to a point, and on an employee by employee basis.
The way backuppc handles it maps pretty well to PCs and single user
workstations but not quite as nicely for multiuser machines. It has a
concept of an 'owner' for each target and a set of administrators. If
you log into the web interface as a non-administrator you can only see
the machines where you are listed as one of the owners. It is possible
to make subsets of a server look like separate machines but it would be
cumbersome for a large number of users if you only wanted them to see
their own home directory.
---
Les Mikesell
lesmikesell@xxxxxxxxx
_______________________________________________
CentOS mailing list
CentOS@xxxxxxxxxx
http://lists.centos.org/mailman/listinfo/centos
