On Fri, 23 Feb 2001, Caffey, Jeff L. wrote:

> REALLY...?   Are you using SAN connection agents or is that performance
> increase still going across your network?

Across the network.  We were looking at IO stats trying to find the
bottleneck as 20 hours for 40Gb was woefull, even for NT, and we found the
server was waiting for the D40 drawer.  We moved the backup pools across
to the shark and it's dramatically faster.

We're now seeing that CPU/network performance is the limiting factor.  The
slowest clients are a couple of P200Pro NT servers, the fastest some H80s.
All across 100BaseT.


Carl.

>  -----Original Message-----
> From:   Carl Makin [mailto:carl AT XENA.IPAUSTRALIA.GOV DOT AU]
> Sent:   Thursday, February 22, 2001 5:46 PM
> To:     ADSM-L AT VM.MARIST DOT EDU
> Subject:        Re: FIVE questions for TSM 4.1.2 (server on AIX, clients on
> Windows)
>
> We moved our 120Gb disk backup pool from a SSA D40 drawer (RAID5) to the
> Shark and got an order of magnitude improvement in backup times.  40Gb
> NOTES backups went from 20 hours to 4 hours.  The shark outperforms the
> SSA disk by a significant amount.  If you start having performance