>> On Thu, 26 Jan 2006 20:24:40 +0100, Dirk Kastens <Dirk.Kastens AT 
>> UNI-OSNABRUECK DOT DE> said:

> That's interesting. I just made some tests on our new server. At first I
> defined three big dbvolumes (using raw logical volumes), each on a
> separate hdisk (AIX 5.3), and each being mirrored by TSM on another
> hdisk. When TSM started up, or when I did an auditdb, only one of the
> primary volumes was active at a time, with a read throughput of about
> 1000 kB/s (measured with topas). Then I defined striped logical volumes
> using three hdisks. During startup or an auditdb all three disks were
> active at a time, but the read throughput of each disk was degraded to
> 250 kB/s, what makes 750 kB/s for the whole volume.
> That means that the throughput was higher with separated volumes. Or did
> I make a mistake?

Your observations are close to mine, but that startup action is not
very similar to any other activity the TSM server will do.


- Allen S. Rout