Hi Bill,

Migration and backup stgpools shall take some extra time as there are more
volumes te mount/dismount.
If you have to migrate "small" amounts during backupperiod because you don't
have enough diskpool, the backup period could be much much longer.

If you define the collocation groups, don't put all the important servers
together in a group, but mix them with less important servers. So if you
have to restore multiple servers, they don't have still all the data
together on the same tapes.

After the define of the groups you can move the nodedata of a group to get
all the data of a group together. move nodedata colloc=....

Databases etc can normaly be grouped all together in 1 group, as they have
their full backups often. So they don't need to be in separate groups. I
usually put DB data in a separate stgpool without collocation, as they don't
need space reclaim when the retention period is the same for all DB clients.

Regards,
Maurice van 't Loo
TSM Freelancer


2010/11/12 Evans, Bill <bevans AT fhcrc DOT org>

> Does anyone have an real world example of moving from a single storage
> pool to collation by filespace?
>
>
>
> TSM 5.5, AIX 5.3
>
> I'm backing up a 200TB server (solaris) with 1TB of new/changed files
> per day.  There are about 30 volumes (filespaces) and I am thinking of
> changing to collation to improve restore times.
>
>
>
> Some things I can figure out, like it will require 30+ minimum scratch
> tapes per night (typically we run with 5).  And each collation group
> will have its own 'filling' tape.  Other than the extra library slots,
> are there any other things I haven't thought of that will increase my
> h/w costs or administration time?
>
>
>
> Thanks,
>
> Bill Evans
>
> Storage and Server Administration
>
> 206-667-4194
>
>
>