Thread started: Jan 19 2010, 9:31 AM EST
Watch
We did some testing and found that the only value that compared to the default value of 1000 was 1. Changing from 1 to 2 dramatically changed the results. If anyone can offer an explanation as to why this is, I would love to understand.
We tested the following values: 1,2,5,10,75,100,250,500,1000,4200.
I will paste our results in a second post, ran out of room on this post. If you copy them to a text doc and change the extension to csv you can have a look.
Based on our testing and the recent bug in 4U1 that changes the value to a random number…tough to recommend changing the IOPS value.
Perhaps the IOMeter load I selected skewed the results. And if you had a well defined load (not random) there might be a value that works better than the default.
out of
found this valuable.
Do you find this valuable?
Show Last Reply
|
|
Last Reply:
RE: iSCSI RR IOPS w/ CX4
By: ,
Jan 19 2010, 10:14 AM EST
We are not using jumbo frames. MTU 1500. The only thing we collected during the runs was esxtop, I have screen shots I can email you. We do not have NaviAnalyzer, so limited visibility as to what was going on in the CX4. The r710 did not show any signifcant bump in CPU.
out of
found this valuable.
Do you find this valuable?
|