ADSM-L

Re: Does subfile backup reduce tape occupancy?

2005-11-04 12:12:36
Subject: Re: Does subfile backup reduce tape occupancy?
From: "Prather, Wanda" <Wanda.Prather AT JHUAPL DOT EDU>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Fri, 4 Nov 2005 12:12:21 -0500
HI Tab,

I think "it depends".

When I was using it, the answer was yes.  What we used it for was
squishing a lot of .pst files.  They tended to be fairly large, up to
500m, and were excellent candidates for subfile, because only a small %
of the .pst file actually did change each day.  (Besides benefits to
occupancy, with a lot of clients in the config it reduces the
requirements for your disk pool, the load on your network, etc.)

With subfile on, and 3 versions, the first backup (of a file) after you
turn on subfile will still be the entire file, because it has to create
the base file in the cache directory.  So don't be surprised that you
get no benefit on the first run  (and don't forget to exclude the
subfile cache from backups!)

The next 2 versions sent from the client will just include the BLOCKS
that change.  And the server just records the number of BYTES that the
client sends to it, so the recorded occupancy will drop, and the actual
occupancy will drop.

However, with only 3 versions, you many not get that large a % of
benefit, because remember that even when you get to the 7th version of a
file, the server is still holding on to the base, plus 6 deltas - there
always has to be a base file in server storage, which is full size.  So
in your case, you are getting rid of less than 2 versions, in terms of
bytes saved.  The benefit would be higher, if you were starting out with
more versions in server storage.

And remember, 
- very small files don't qualify for subfile
- very LARGE files don't qualify for subfile
- if a large enough percentage of the file changes, it does a full (of
that file) anyway.

So I have no doubt that you can construct cases, where the answer is YES
or the answer is NO, depending on the data. 

On the other hand, it's a harmless thing to test out.  Turn it on for
just one of your servers, and monitor your occupancy.  You should have a
good idea of the impact in a week or so!

Caveat:  DON'T EVEN THINK ABOUT IT, unless you are using collocation!  A
restore requires base plus deltas - you don't want to do multiple  tape
mounts to get 1 FILE back!

Wanda Prather
"I/O, I/O, It's all about I/O"  -(me)




-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
Tab Trepagnier
Sent: Friday, November 04, 2005 9:55 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Does subfile backup reduce tape occupancy?


Like lots of users, our tape libraries are overloaded and we're looking
for ways to reduce what we're keeping before spending money on new
equipment.

Server: TSM 5.1.10 on AIX 5.2
Clients: mostly TSM 5.1.7 on Windows XP/200x.
Storage: mix of disk, 3570, and LTO
Most filesystems are configured to maintain three versions.

If I have a server whose current occupancy is 1000 GB, and I turn on
subfile backups with the same number of versions, will the actual
occupancy go down?  Or will it remain the same?

Thanks.

Tab Trepagnier
TSM Administrator
Laitram, L.L.C.

<Prev in Thread] Current Thread [Next in Thread>