VMware Cloud Community
carlewarl
Contributor
Contributor

Disk latency

Hi All

I'm wondering if someone can help me with the following.

We're having what would appear to be general storage latency issues in our VMware environment.

Picking one VM that has been reported to be performing badly I can see in Ops Manager a very large spike in write latency to the virtual disk but this is not reflected in the write latency to the datastore - see the attached screenshot.

Can anyone explain where this latency might be coming from please?

Thanks.


Carl

Reply
0 Kudos
2 Replies
daphnissov
Immortal
Immortal

You're somewhat comparing apples to oranges. In the first two graphs, you're looking at disk I/O latency inside the guest and at the datastore level, respectively. The third graph is a different graph which shows SCSI command latency, not I/O latency. But if you look closely, the yellow dots indicate observed highs and lows. The blue shaded graph underneath shows the profile of that line graph. When you look closely, they all align. So what's probably happening is the datastore I/O latency (whatever happened there, don't know) caused SCSI command latency to increase which caused virtual disk I/O to increase. This could have been a network/fabric event, a disk device/extent issue on the array, or something else. But the profiles align to tell the same effective story.

Reply
0 Kudos
carlewarl
Contributor
Contributor

Thanks for your reply daphnissov.

Reply
0 Kudos