FR: Hyper-V Guest Uptime calculation change

[nomis35]

As per case ref: 02417524 - We noticed that the VM uptime statistics in Hyper-V are incorrect, due to the way the statistics are collected:

Technical Support gave the following response when we raised this:

By design, VeeamOne collects the VM`s uptime information from the Hypervisor and by Microsoft design, it is the usual situation when Guest OS uptime has a difference with Hyper-V uptime metric because each time you move the VM from one host to another this counter drops.

I would like to suggest that the VM uptime is obtained from the Guest rather than the host, as in the case of VM's frequently moving, the statistics are incorrect. Am I right in thinking that the chargeback report uses the VM uptime statistic to calculate charges, and will hence be incorrect?

[Shestakov]

Hello Simon and welcome to the community.
That looks like a bad design. We are working with support and developers to find out if we can improve the behavior.
As for Chargeback reports, uptime there is just for information. Costs are calculated based on the performance during the interval.
Thanks for the heads up!

[nomis35]

Thanks for the reply - will look forward to possible changes in the future.

As our proof-of-concept install has expired its license I cant check for definite, but I'm sure when we looked at charge back, two similarly configured servers with equal performance, but with different up times reported, were giving different costs for charge back - we know the up times were incorrect for one server as it was monitored by our SCOM system.

[nomis35]

Furthermore, looking at the documentation here:

https://helpcenter.veeam.com/docs/one/r ... tml?ver=95

For the VM performance Chargeback report,

The cost of VM resources is calculated based on VM uptime multipled by the cost of average CPU, memory and storage usage observed during a specified period.

[Shestakov]

In this case we need to slightly change the description.
The report operates with performance counters, which implicitly include uptime.
2 days with 40% of CPU usage will give same cost as 80% for the first day and power-off i.e. 0% for the second day.
Thanks!

[nomis35]

So that surely means that if the uptime is incorrect due to the inaccurate statistics from the host, then that particular chargeback report is essentially useless as it will also be incorrect - that's a major issue to anyone using Hyper-V who wants to do chargeback?

[vfiler]

A new support case was opened at the request of Veeam: #03204344 - Issue with VM uptime statistics in Hyper-V in relation to the above case.

We are still awaiting a fix which will allow everyone who uses this product in a Hyper-V environment to collect and report upon the actual uptime of VM's rather than the metric that it currently uses which gets reset when VM's are migrated or any other actions which create a change.

We have been told that the Feature Request is with the R&D department but we would like a timeframe for the implementation of this fix so that we can utilise reporting correctly.

[Shestakov]

Hi vfiler,
Upcoming release will have the updated version of VM Uptime report for Hyper-V machines.
Besides requested downtime in seconds/minutes, it will contain details about VM restarts and related alarms.
The release is coming later this year. You may ask Veeam system engineer to deliver you beta version of the release.
Thanks!

[Shestakov]

Hello,
Update 4 is released and available. It contains the updated version of VM Uptime report and a number of other great features.