Monitoring and reporting for Veeam Backup & Replication, VMware vSphere and Microsoft Hyper-V in a single System Center Operations Manager Console
Post Reply
Posts: 225
Liked: 13 times
Joined: Feb 14, 2012 8:56 pm
Full Name: Collin P

VM Latency Alerts

Post by collinp »

In the latest v7 Hyper-v management pack, I get alerts on CSV latency. It would be nice if I can drill down and see which VM is most responsible for this latency. I can see IOPS per VM but I cannot see latency per VM. In the Windows Server counters I can see that there is a latency counter but it would nice if the Veeam Management Pack had a way to show which VM(s) were most responsible for the latency on the CSV.
Posts: 452
Liked: 76 times
Joined: May 02, 2012 1:49 pm
Full Name: Sergey Goncharenko

Re: VM Latency Alerts

Post by sergey.g »

Hello Collin,

You should be able to open Disk Performance Analysis dashabords right in the context of the alert, here you will see IOPS for VMs. You are correct - we don't have disk latency for each VM and this is a good feature request for our next versions. But in reality, latency for VMs is not a good indication of the source of the performance issue. When a Disk experience high latency, then all VMs on this disk is going to suffer from that and even a VM which is doing almost nothing will have to wait "in queue" to get its commands completed, so high latency of a virtual machine does not necessarily mean that this machine is responsible for traffic storm.

We would recommend to check VMs which are producing most IOPS, usually this is a Virtual Machine responsible for performance degradation: faster and in more quantities it's sending commands - less time the disk has to complete requests from other VMs.

Also if you have a SCOM agent inside guest OS, you can check Disk Time counter (it should be visible if you open a Performance view for the disk, because all VMs and the corresponding Windows Computers have relationship with the disk). This counter represents virtual disk queue size of the OS, which is going to be higher for a virtual machine which has most traffic (even if some of this traffic queued and on the hypervisor level IOPS counter is relatively the same for several VMs)

Thank you very much for your interest in our product and for your feedback and feature request. Let me know if you have any other questions.
Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest