Re: [ADSM-L] Active Storage Pools
2013-11-08 07:15:12
Hi Nick
It sounds like you are between a rock and a hard place. Are the lawyers
really allowed to obstruct normal operations like that? Not reclaiming
will cause the number of tapes required for a complete restore of a
client to blow out to the point where recovery time would be
impractical. Its a bit like telling a restaurant that they can't throw
out their garbage. Eventually the place will become unsanitary and have
to close and that would be an abuse of process.
As to expire inventory, run it. Don't forget that even if you do not
run expiration you still cannot get at those objects. AFAIK they are
still expired, but they have not been cleaned up yet. So you might as
well run expiration. You will need to anyway to make the active pool
solution work.
To populate your active pool: rename your existing primary pool and
create a new one with the old name. Run a COPY ACTIVEDATA against the
new pool daily and you will get only new data into the active pool.
If desired, you can catch up on the pre-existing active files by running
selective backups on each node as resources permit.
HTH
Steve
Steven Harris
TSM Admin
Canberra Australia
On 8/11/2013 10:41 PM, Nick Laflamme wrote:
Does anyone have much experience with active storage pools?
My current customer isn’t allowed to reclaim tapes, so their inventory of
primary storage volumes is much larger than their ATL. This makes client restores
prone to failure because of the number of primary volumes that need manual
intervention as well as the lack of routine operational support at the site.
I’m considering creating an active storage pool. Once fully populated, its tapes would remain in the ATL; mere primary volumes would be
on-site but not in the ATL. I know I can start to create it by defining an active data pool to be written to as primary tapes are being
written to (we already use this for copy volumes), but there’s still the matter of copying into the active pool the active data on the
existing primary storage volumes. Is there any way to build the active storage pool in phases, such as “all the data that’s four
weeks old or younger” or “all the data on volumes that are available; don’t flinch at volumes that don’t
mount”? Otherwise, the COPY ACTIVEDATA process is going to run 24x7 for a long, long time. If I cancel an COPY ACTIVEDATA command and
then start a new one, does it correctly understand what it no longer has to copy?
Do reclaims of active data storage pools rely upon EXPIRE INVENTORY running regularly? We
haven’t been running EXPIRE INVENTORY out of an abundance of caution to avoid any risk
that we lose track of an inactive object someone might demand from us, but I suspect
we’ll need EXPIRE INVENTORY to keep the active pool correctly populated. I can imagine
how to set the retention policies to mimic not running EXPIRE INVENTORY, but that still might
make Some People nervous.
Thanks to Wanda’s tip about EXPORT NODE a couple of weeks ago, I think I know
how large (roughly) my active data pool would be if I create it and populate it.
What am I forgetting? Is my plan fatally flawed?
TSM Server 6.3.4, if it matters.
Thanks,
Nick
|
|
|