ADSM-L

Re: [ADSM-L] TSM server scaling/sizing for lots (>20000) nodes

2008-04-29 09:17:11
Subject: Re: [ADSM-L] TSM server scaling/sizing for lots (>20000) nodes
From: Zoltan Forray/AC/VCU <zforray AT VCU DOT EDU>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Tue, 29 Apr 2008 09:15:50 -0400
I thought it was 800-bytes for each entry?

We will continue to expand our 3494 library. We can easily add a rack of
12-3592 drives.  All it takes is $$$$$$$$$$$$$$$$$$$$$$.



Remco Post <r.post AT PLCS DOT NL>
Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
04/28/2008 05:51 PM
Please respond to
"ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>


To
ADSM-L AT VM.MARIST DOT EDU
cc

Subject
Re: [ADSM-L] TSM server scaling/sizing for lots (>20000) nodes






Zoltan Forray/AC/VCU wrote:
> Crazy project time...........
>
> I have been asked to research the
> feasability/probability/scaleability/sizing of TSM to possibly handle
> backing up thousands (actually close to 20,000 or more) desktops!
>
> Anybody have experience doing very, very large scale backups of so many
> nodes?   Don't expect much in storage occupancy since we would exclude
> things like MP3, etc.  It would just be the ##### of files/database size
> (sounds like this would have to wait for V6 since DB2 should scale much
> better).
>
> Does anyone run their TSM servers, open?   Trying to
> administer/register/monitor this many clients would be a nightmare.
>
> They are talking about budgeting close to $1M, which would include more
> bodies as well as the equipment/software!

Well, you are an experienced TSM admin, so I guess you know the drill,
about 500 bytes for each version of each file. Get a rough estimate of
the number of files. The number of nodes you mention is an issue, even
if each stores just a few files (they won't I guess).

You'll need lots of TSM instances, and lots of db spindles. Really,
there is a use for 15k RPM 72 GB  (or even 36 GB) disks in raid1, and
you just found one.

In my experience, there are a few things to make life easier, Gresham
EDT is one of them, with autolabeling and decent library sharing, you
really gain over TSM's native library sharing. On the plus-side, you are
limited to ACSLS libraries, which you'll probably need anyway for larger
amounts of data (how many options are there, really? ;-) ).

--

Met vriendelijke groeten,

Remco Post

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