Hi, I'm getting poor performance from a qla2340 adapter attached to a 2GB switch and an EVA3000 behind it, so I hoped to improve it by pulling from both active paths. The host is RHEL4U2 dual opteron using 0.4.5 multipath tools. I used group_by_serial and had some improvement, but less than I hoped. I remember that there were some performance issues with group_by_serial and storageworks tagged as 20% loss. Is this still true? A bit off-topic, but I'm trying to benchmark the throughput. Does anyone have numbers for qla2340/brocade 2GB/HSV100 or a pointer? Thanks! -- Axel.Thimm at ATrpms.net
Attachment:
pgp3MEkefwGKi.pgp
Description: PGP signature