I have three identical nodes (basic Lenovo servers) but one performs really slowly compared to the others. The disks aren’t the same across all three though. I wonder if anyone has any ideas which difference might be causing the performance issues:
Firstly the HDDs are 4TB in the two ‘good’ nodes and 2TB in the ‘slower’ one. I wouldn’t expect that to cause performance issues, just overall capacity will be limited? To be fair the 2TB is a WD ‘green’ and the 4TBs are WD ‘red’ drives.
Secondly the ‘slower’ host has a Samsung Pro 840 256GB SSD which Prism reports as often having a higher latency than the HDDs! See the image below
Are they known for being poor in this kind of environment? I’m not really stressing them. Any good logs to see performance issues? I can see the HDD activity light is permanently on on the slower host.
Best answer by SteveCoopsView original
Here is a good writeup on disk performances: https://next.nutanix.com/how-it-works-22/disk-i-o-latency-on-a-nutanix-cluster-38349
Forgot to update this post - after a couple of days this problem went away. Seems a common thing on some Samsung SSDs that they’re slow for a while after a format.
Ok thanks for clearing things up. Good luck!