Some heads up on an important change we’ve been working with VMware that is now part of the vSphere 6.0/6.5 and 5.5P8
If you attach an ESXi host to an XtremIO array, it will AUTOMATICALLY choose Round Robin (RR) and IOPS=1 as the SATP behavior.
a little background.. because XtremIO is an Active/Active array (all the storage controllers are sharing the performance & ownership of the entire data), the default ESXi SATP behaviour was set to “fixed” which means, only one path for every datastore was actually working, of course, you could have changed it using the ESXi UI, using the Dell EMC VSI plugin etc..
One word of cautious, This does not mean that you can skip the OTHER ESXI host best practices which you can either achieve using the VSI plugin as show below (HBA Qdepth etc) or by using the ESX CLI, powershell, scripts etc
But it does mean that if a customer didn’t read the HCG or forgot about it, many errors we have seen in the field won’t be seen again due to multipathing misconfiguration. It will also mean much better performance out of the box!
For ESXi 5.5, you must install the latest update (https://kb.vmware.com/selfservice/microsites/search.do?cmd=displayKC&docType=kc&externalId=2144361&sliceId=1&docTypeID=DT_KB_1_1&dialogID=298460838&stateId=1%200%20298476103 )
For ESXi 6.0, you must install the latest update – 2145663 (https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2145664 )
For ESXi 6.5, it’s there at the GA build (which is the only build for now)
If you want to run a query to validate this:
|# esxcli storage nmp device list|
Lastly, the latest VSI plugin (7.0) doesn’t support vSphere 6.5 yet, the VSI team are working on it and will update once I have more news.