So I have a ML310e Gen8 v2 with 4 Samsung 845 DC EVO SSD (RAID10) runing ESXI 6.0.
I've 4 VMs and I run a crystalmark benchmark on a windows VM.
Only got 100MB/s read and write MAX.
Don't know if the issue is the onboard "raid controller" (b120i) or something with ESXI...
I believe that I should get higher benchs.
Thanks
Can you check VMware HCl whether the controller is supported one and check for what the max iops supported.
From what I managed to understand it's supported.
91430 IOPs
HI,
same problem here, after upgrade from esxi 5.1 to esxi 6 slow local datastore performance, proven on 3 different systems, 2xML310e Gen8 v2 and 1xMicroserver Gen8 with different types of SSD (Intel/Samsung) and HDD also.
NNow that you are saying that i remember that I had better performance with 5.5
Ok,
The issue was HPVSA vib.
Basically you'll need to:
Uninstall latest VIB
esxcli software vib remove -n Hewlett-Packard:scsi-hpvsa
Install older VIB (84 or 88)
esxcli software vib install --viburl=/vmfs/volumes/DATASTORE/scsi-hpvsa-5.5.0-88OEM.550.0.0.1331820.x86_64.vib
Benchmarks now:
518MB/s Read | 215MB/s Write.
In any case I bought a P420 controller. I'll let you know the results as soon as I install it.
Thanks!
Worked as promised!
Thank you!
BR
Aba
Just wanted to add my thanks to the poster of this solution. Applied this fix in the lab earlier and performance is much better.
Cheers!