REAL LIFE performance with TSM Server 6.4 and IBM 7650G (single node & DS4700)

[Rigido]

Hi Admins,
sorry for the long title but I wasted many hours searching online to find "case studies" of people with unlimited resources.
Actually I have:

• IBM 7650G + DS4700 (16 x 1TB disks)
• HP server with Windows 2008-R2 64bit, TSM 6.3.3.0, direct attached to the 7650G (2 fiber, one per controller) and "Fabric" attached to Netapp Storage (2 fiber connected)
• On the TSM server I have configured the TDP for VE (SAN access to snapshots on the Netapp storage)

After some server and client tuning (using sharedmem) now we are backing up a single 200GB virtual machine at 40MB/sec (as reported by the ProtectTIER Manager),
then performance degraded to 2MB/sec.
Tried to run multiple sessions (we have a VMware cluster with 5 servers, TDP for VE configured to use 5 sessions max 1 per server) started at 100MB/sec then dropped to 40MB/sec (5 x 8MB/sec).

Any hints?

 

[Rigido]

This quick post Just to let you know that the performance issues I was talking about disappeared.
Probably the VTL was stuck in some sort of defrag/dedup/reorg process as we deleted about 200 volumes, modified VTL settings and defined 300 brand new LTO volumes (40GB each).
We did some tests and now we are at a fixed throughput of 100MB/sec.