I believe that the restore performance for small files is impacted by
how spread out they are on tape (the same tape, not different
tapes). If all of the files you are restoring are contiguous, I
believe that tape can perform much faster. The problem we run into
is that there are gaps between the files. Caused either by expired
files, or other nodes sharing the same tape (even with collocation,
if you have multiple nodes per tape). Thus, when doing performance
testing on tape, it is important to understand the test data that you
are restoring (i.e., whether the files are contiguous on tape or
not). We have found this makes a noticeable difference.
Jon, when you state that "the more clients you add, the slower each
backup stream becomes", are you talking about TSM clients? If so,
are you talking about simultaneous backup streams, or just the fact
that multiple clients are registered on the TSM server? I'm trying
to understand what the cause of the performance degradation is.
We have also been considering data reduction type devices to add to
our TSM environment. I am curious whether you are using client
compression or client encryption, and to what degree that interferes
with the reduction. FYI, Sepaton recently announced a data reduction
capability on their vTape product. I don't know too much about it,
but I think it does after-the-fact reduction by scanning files already stored.
..Paul
At 10:49 AM 5/16/2006, Jon Evans wrote:
I have recently tested a DD460 for exactly the same reason. The results
showed that compression was good (upto 20x) but throughput was not so
good. The more clients you add, the slower each backup stream becomes.
Off course, much depends on your infrastructure, and these results were
based on a windows filesystem, running across a gb network.
Unfortunately, with millions of small files it is very difficult to
improve performance significantly.
Personally, I could not improve backup or restore performance over an
LTO2 or LTO3 tape drive, and the tape drives still work out cheaper.. so
I decided against it at this time..
Regards
Jon Evans
-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
Christoph Pilgram
Sent: 16 May 2006 14:43
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Disk-to-Disk Backup
Hi all,
Because we have problems to hold our service level agreements with the
customers for restoring big file-servers (10 Mio files, 1TB disk-space
in one filesystem), we are thinking about storing the backups not
anymore on tape but on disk. Does anybody has experience with that kind
of storage-pool for about 40 TB of backup data ? Does anybody use for
example a "Data Domain DD460" or other systems using COS to reduce the
amount of data.
Thanks for help
Chris
----------------------------------------------------------------------
This e-mail, including any attached files, may contain confidential
and privileged information for the sole use of the intended
recipient. Any review, use, distribution, or disclosure by others
is strictly prohibited. If you are not the intended recipient (or
authorized to receive information for the intended recipient),
please contact the sender by reply e-mail and delete all copies of
this message.
--
Paul Zarnowski Ph: 607-255-4757
Manager, Storage Systems Fx: 607-255-8521
719 Rhodes Hall, Ithaca, NY 14853-3801 Em: psz1 AT cornell DOT edu
|
