Directly attaching the iSCSI was a first step measure to rule out major switch problems. Setting up MPIO after I hit the limit of the 1 GB HBA was my next step.
I mixed up my terminology, I meant outstanding IOs were bumped up to 8, not workers.
I have been watching SANHQ, which is why I do think there is a problem: http://i.imgur.com/2q62i.png. This is on a production SAN. As you can see, as IOPS rarely go over 2000, latency is often high, and I've never seen a transfer rate over 100 MB/s.
lcasteel
1 Rookie
•
30 Posts
0
June 1st, 2012 14:00
Edit: Meant Raid50 on each member.
lcasteel
1 Rookie
•
30 Posts
0
June 1st, 2012 15:00
Directly attaching the iSCSI was a first step measure to rule out major switch problems. Setting up MPIO after I hit the limit of the 1 GB HBA was my next step.
I mixed up my terminology, I meant outstanding IOs were bumped up to 8, not workers.
I have been watching SANHQ, which is why I do think there is a problem: http://i.imgur.com/2q62i.png. This is on a production SAN. As you can see, as IOPS rarely go over 2000, latency is often high, and I've never seen a transfer rate over 100 MB/s.
Thanks for the info, this gives me some ideas.