Just after some advice. I’ve run the default ‘Four Corners Microbenchmark’ against 2 x identical All Flash clusters and I get very different results and not sure why.
The average random read IOPS differ by around 60k, and the random write IOPS around 10k. I know running the test on both clusters will not provide an exact match, but these results appear very different. Please note these are both 2-node clusters, both are identical in specification so not sure if this matters.
I also get the same on 2 x identical 6-node Hybrid arrays, random read IOPS differ by 200k and random write IOPS by 40k!
The same X-Ray appliance was used on all clusters, using the default settings in the test itself.
Just after a possible explanation to these queries, if somebody could be so helpful.
Thanks very much.
Best answer by AlonaView original
I suggest raising this with a Nutanix SE. Let me know if you have contact details or if you need help with getting in touch with one.