ADSM-L

Re: [ADSM-L] Re: Policy for NDMP dumps

2011-05-04 17:26:06
Subject: Re: [ADSM-L] Re: Policy for NDMP dumps
From: David Bronder <david-bronder AT UIOWA DOT EDU>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Wed, 4 May 2011 16:21:12 -0500
Gee, Norman wrote:
>
> You can assign a different management class to full and differentials.
> I have defined them to different tape storage pools.

That would work for storing the different NDMP images in different
storage pools.  But my experience matches Remco's: each "backup node"
command that specifies a different management class will rebind all
backups (full or differential) for that node/filespace, including
adjusting the retention.

I finally gave up trying to use longer retention for full NDMP images
(e.g., a year) and shorter retention for differential images (e.g.,
four weeks).  Though I never tried Remco's suggestion of sending the
differentials to a different node name with different policy.

I have nothing good to say about NDMP.  It had its place 15 or 20 years
ago, but not today.  I'm really hoping I'll be able to replace NDMP for
our NetApp backups with SnapDiff-based backups.  That should save time
and space, should generally perform better, and of particular interest
to me, should get rid of long-running single-object transactions that
pin the recovery log for insane lengths of time (inevitably filling the
recovery log to 90+% and eventually auto-canceling the NDMP job just
before if finally completes, wasting 20+ hours of time and terabytes of
bandwidth plus sometimes dozens of tapes for TSM DB backups).

(Unfortunately for Wanda, SnapDiff is NetApp/N-series only, so it won't
help her Celerra backup situation.)


--
Hello World.                                    David Bronder - Systems Admin
Segmentation Fault                                      ITS-EI, Univ. of Iowa
Core dumped, disk trashed, quota filled, soda warm.   david-bronder AT uiowa 
DOT edu

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