Is there a threshold where cluster latency would affect VM performance?
Page 1 / 1
In traditional storage probably anything steady over 20 ms can be a cause of concern due to hardware dependcy filling up like NVRAM. With Nutanix we don't have any reliance on NVRAM so performance scales as nodes get added and is usally pretty steady.
I've seen < 30 ms just work fine on Nutanix(spikes). Some of the latency also may appear higher as secondary health checks happen on the cluster. It's really a depends answer because if an appliocation is using a larger block size, latency could go up, but throughput maybe be really good.
If you're having an issue contact support away. Also running the NCC(Nutanix Cluster Check) can give you some peice of mind.
I've seen < 30 ms just work fine on Nutanix(spikes). Some of the latency also may appear higher as secondary health checks happen on the cluster. It's really a depends answer because if an appliocation is using a larger block size, latency could go up, but throughput maybe be really good.
If you're having an issue contact support away. Also running the NCC(Nutanix Cluster Check) can give you some peice of mind.
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.