ADSM-L

[ADSM-L] Need guidance on setup of multiple replicated/protected+tape-copy dir container pools

2017-01-31 08:58:07
Subject: [ADSM-L] Need guidance on setup of multiple replicated/protected+tape-copy dir container pools
From: "Michaud, Luc [Analyste principal - environnement AIX]" <Luc.Michaud2 AT STM DOT INFO>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Tue, 31 Jan 2017 08:56:52 -0500
Hello gents,  I need your guidance on the setup of multiple 
replicated/protected/tape-copied directory container pools.

First, a bit of context :

1.      We are in the early phases of going from client-dedup'ed seqpools, to 
inline dir-cont-pools.

2.      The good news is that the data owner is ok with going from scratch on 
the new cont-pools, so no need to convert any of the existing node information.

3.      The replicated cont-pools are housed on 2 distinct TSM 7.1.7 instances 
on AIX 7.1.3, built from scratch using the medium blueprint.

4.      Even though we've just added replicate+protect to the mix, we still 
need to copy-to-tape - until we get our act together on 
copy-to-cloud-containers (mostly internal procureme
nt issues with cloud).

5.      For the first bunch - a bunch of Windows 2012 clients doing CIFS File 
Services, the results were impressive, We intaked 50TB client-side capacity in 
a hurry, and copy-to-tape is
blazing fast.  Are we ever happy !!!


NOW we need to plan to get rid of the other dedup'ed seqpools.  This brings a 
bunch of questions, to which IBM support won't really meddle into (yep, I 
tried), and that I have to convince peers :

1.      Should I create lots of smaller pools whenever possible ?  For example, 
the OS-SEQPOOL here contains AIX, Windows and LINUX nodes.  I think I ought to 
split it into 3 different OS
-specific pools, as they don't dedup much of each other IMHO, and that the 
tape-copy of containers does scale very well for recoveries.

2.      Should I create lots of TSM servers, each with its distinct DB2 ?  Up 
til recently, all the nodes and all the dedup-seqpools were hosted on a LONE 
TSM 7.1.7 server.  To this day,
we can't seem to fix issues causing the *OFFLINE* reorg of dedup indices from 
taking over a DAY, even though we are under 100TB of licensed capacity.

3.      Would you see any issue with the idea of creating a mix of 
source-replicated and target-replicated dir-cont-pools onto each of my 2 
"blueprint" TSM servers ?

All advises are welcome; I will summarize.

LUC MICHAUD, B.Ing
Analyste Principal
Technologies de l'information et innovation
Sociéte de Transport de Montréal

<Prev in Thread] Current Thread [Next in Thread>
  • [ADSM-L] Need guidance on setup of multiple replicated/protected+tape-copy dir container pools, Michaud, Luc [Analyste principal - environnement AIX] <=