S2D gurus, performance again.

[Jeff Robertson]

While one node was down?
And half the speed when it needs to sync over network then?

It seemed to fluctuate between 2-3x the performance at 4K QD1 when one of the nodes was down, everything else was pretty similar. I'm not sure what that means exactly, I've done every test I can think of and the network stack seems to be working as it should. Maybe that is normal?

 

[Rand__]

Well o/c you have extra latency over network so if its limited by waiting for remote sync then it might be, but I am no S2D expert. Just reading this since I am still looking for options

 

[Jeff Robertson]

Yeah I would expect the behavior I'm seeing to some extent, maybe it's spot on, no idea! I ran a 2016 cluster for a couple of years with lower performance and wasn't unhappy with it, VMs were snappy. The only time I felt it was when patching half the VMs at once, it *seemed* to run a bit slow but since I never ran benchmarks it could just be in my head. I did run crystaldisk on two VMs at once and neither seemed to have any performance degradation so maybe I'm seeing normal performance.