Usage Metering - Virtual Machines & Computer Standby
After some questions from a customer, I started investigating some raw usage data and found considerable gaps.
With gaps I mean the number of days since the last usage data delivery. I would usually hope for a weekly consolidation of usage. Admitted, there are examples of that in here (7 & 8), but there are also larger gaps, like 77 days and 85 days.
This is happening to virtual machines and computers. Gaps on VMs seem to be somewhat higher than on hardware. My assumption is that the agent might not track applicaiton usage sessions when
- a VM or Virtual desktop and applications on it are not shut down, or
- a computer/laptop is just sent to standby.
@ChrisG How do you think would the agent behave in these situations? Would it only send usage data when the machine and therefore the app is closed, e.g. forcibly after updating or something?
This thread has been automatically locked due to inactivity.
To continue the discussion, please start a new thread.
Is the question you are wondering about here why there are no rows of data for the weeks starting 2002-05-11, 2005-05-25, 2006-06-08 to 08-17, 2009-08-31 and 2009-09-21?
I can imagine this may be because the usage agent running on the computer didn't upload usage data during those weeks for some reason. That might be because the computer was sleeping or suspended at the time the usage agent would have otherwise uploaded data, or could be for reasons I couldn't guess. Taking a look at usage agent logs covering periods in question may give some insight into the actual behaviors going on.
The thing is that these machines do upload a proper inventory once a day. So they do not seem to sleep/be suspended over longer periods. I guess virtual desktops might just run 24/7. Local VMs might just be suspended/saved. And notebooks might just be closed/sent to sleep.
I'll try to check the logs.
We had issues that employees shutdown their computers before usage agent closed it's session, this ended up in that we got really poor usage follow-up.
I have no stats for this, but the following tweaks we have on our agents. For Windows you have the addremove=true as well.
This looks really promising. I'll ask my customer test this.
But I am still sceptical about machines not closing some apps at all.
Repaired usage time for Windows devices suspended over weekends
Previously, if a Windows device was in a suspended state at 12am on Monday mornings, the report of weekly usage times could not be generated and uploaded, with the result that the previous figures rolled forward into the next reporting period and usage appeared to grow continually. This has now been repaired, and (when usage tracking is enabled) FlexNet inventory agent now triggers the weekly report as soon as the suspended device resumes normal operations. This is particularly valuable for enterprises wanting to use usage as a way to control internal cross-charging for computer use.