Hi... more 2cents.


My admin schedules look like:

3am   COPY-BACKUPPOOL
   cmd backup stgp backuppool copypool

To make a copy of all the data that is on the disk storage pool.



4am   MIGRATE-START
    update stgp backuppool low=10 high=10

To kick of a migration to copy the data from the disk pool to my tape
pool.  This empties out the pool in preperation for the next night's
backup.

I do not put the thresholds down to zero.  I have a lot of clients
that do not back up much data most nights (a few MB each).  To
completely empty out the staging pool takes a LOT of tape mounts.

In addition if there are any backups still running at 4am very nasty
things happen.

I have seen it where migrations continiously kick off because one
client is still backing up to the storage pool.  The data can be
migrated off to tape faster than it arrives.

It is easy to see it becoming pathologically bad for the tape drives
with (say) three clients doing this, two tape drives and a collocated
storage pool.



4:15am   COPY-DLTPOOL
   backup stgp dltpool copypool

Kick of a copy of the DLT tapes.  This waits for the migrations to
complete, then backs up the DLT volumes that had data migrated during
the backup run.


5am       COPY-VOYAGER
  backup stgp voyager copypool

There is a special, high rotation storage pool in the the tape
library.  Again, this kicks off and waits until the migration and
previous copy completes before making the offsite copy.


6am  MIGRATE-END
  update stgp backuppool low=50 high=70

Resets the storage pool to normal thresholds.



The nett effect of all that, is that I should end up with two copies
of most of the data at the end of each night.

And the staging pool is emptied so that the next night's backup does
not kick off a migration for a long time.

And the amount of data read from the tapes is reduced.  Doing the
migration first and the copy second means it would re-read the
migrated data to make a copy of it.  As Dan says.



What puzzled me is why ADSM will not use the cached copy in the disk
pool when backing up the files.


Russell


[Charset iso-8859-1 unsupported, filtering to ASCII...]
> Steven,
>
> As others have suggested, the best method to protect your server from
loss
> of data is to implement a strategy to backup your storage pools to a copy
> pool on a daily basis.  One thing that I did not see mentioned is the
> potential performance impact on nightly backup when migration (to tape)
runs
> concurrently with the nightly backups.  The impact can be significant,
> especially on "busy" machine.  I normally implement an administrative
> schedule to migrate all the data on disk pools to tape on a daily basis.
> You can do this by setting the hi threshold of the disk storage pool to
0:
>
>         update stg STORAGEPOOLNAME hi=0 lo=0
>
> You will also need to setup a schedule to reset the parameters to their
> normal values after the disk has been cleared:
>
>         update stg STORAGEPOOLNAME hi=80 lo=50
>
> If you also choose to implement a schedule to backup the disk storage
> pool(s) on a daily basis, you should schedule it to run before you clear
> your disk storage pools to minimize tape mounts.  Implementing both of
these
> strategies should provide you with better data protection and
performance.
>
> DanC
> =============================================
> Daniel G. Crouse
> IBM Certified Specialist - ADSM
> Mainstar Software Corporation
> Storage and Recovery Solutions
>   URL:  <www.mainstar.com>
> Email:  Dan.Crouse AT mainstar DOT com <mailto:Dan.Crouse AT Mainstar DOT com>
> Vmail:  800-233-6838
> Voice:  404-714-5754
>    Fax:  770-682-7739
> =============================================
>
> > > -----Original Message-----
> > > From:       Steven McNicholas [SMTP:ste_mc AT ZDNETMAIL DOT COM]
> > > Sent:       Friday, November 20, 1998 8:52 AM
> > > To: ADSM-L AT VM.MARIST DOT EDU
> > > Subject:    A question about migration
> > >
> > > Hi,
> > >
> > > In the company I am working for their adsm system is used in the
> following
> > > way....
> > >
> > > 1) aix clients are backup up to disk pools
> > > 2) These disk pools have migration thresholds of
> > > low=0% high  %
> > > 3) If the high is reached then it is migrated to a tape pool.
> > >
> > > My question is this :
> > >
> > > If during the daily backup this high threshold is not met on some of
the
> > > disk pools,then this means that there is a whole load of client data
> > > residing on (in our case)  1 disk.
> > > If that disk pops then goodbye data ( could be maybe 3 days worth ?)
> > >
> > > What is the solution to this ?
> > >
> > > Is it possible to mirror the disk pools aix logical volume  to
another
> > > disk using aix ? or is this not possible because there are adsm disk
> pools
> > > in the lv ?.
> > >
> > > Would it just be better to lower the high migrate threshold to a very
> low
> > > value ?
> > >
> > > There is no disaster recovery as yet for this machine (no offsite
> backups
> > > etc ) but in the meantime I still do not want to end up losing any
data
> > > due to disk failures etc.
> > >
> > > Please help.
> > >
> > >
> > >
> > > Free web-based email, anytime, anywhere!
> > > ZDNet Mail - http://www.zdnetmail.com