High NL disk bandwidth for unknown reason

[Lupu]

Small F200 system here housing 32 x 300GB FC 15k and 16 x 2TB NL 7K disks with AO, DO, VC, SR, Thin prov+conv+recl+pers and InServ version 3.1.1 (MU1) P13.

Something strange happened recently. Performance stalled without apparent reason. Looking at SR graphs (attached) we could see that at the same time:

- Average bandwidth per NL disk jumped from the usual 70 Kbytes/s to 10 Mbytes/s
- Total disk port bandwidth jumped from the usual 100 Mbytes/s to 220 Mbytes/s
- Total host VLUN bandwidth dropped from the usual 30 Mbytes/s to 6 Mbytes/s

So it seems something caused large amount of NL disk bandwidth but it wasn't any of our hosts. We checked every host and not much was happening there because of too high service times.

There were no tasks like AO running in InServer at the moment. Unfortunately I wasn't able use Performance Analyzer in SP during the issue. After almost two hours suddenly the issue resolved and bandwidth/performance went back to normal.

Any idea what could have caused this? Thanks!

[Richard Siemers]

Possible it was compacting your NL cpgs? I believe there is an auto option for that.