IOmeter accuracy

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

IOmeter accuracy

perfuser123
hi @all,

testing iSCSI with datacore sanmeledoy with 30 gb lun mapped to a native windows server 2008 r2 sp1 (lun is not initialized on windows - its still a raw device).
using following access specs:
<sequential><random><write><read>
s100r0w100r0
s100r0w0r100
s50r50w50r50
s0r100w75r25
s0r100w50r50
s0r100w25r75
s0r100w0r100

4 workers each worker doing all 7 specs 5 min long each with 1 Outstanding IOs with 4 kB blocksize.

first run result of s100r0w100r0:14200.97 total iops
second run result of s100r0w100r0: 16134.24 total iops

Are these values comparable? how much difference can be ? to say ist okay.
I mean if the first run result is 14200.97 total iops and the second 100000.00 total iops, than I would say something is wrong... but so....

cann anyone help, or tell my what difference is acceptable for the runs?
need help fast...

I am from germany, so dont wonder about my english skills



Loading...