We have a XenDesktop 5 Citrix VDI system, sat upon VMware vSphere 5 system.
I have identified datastore usage alarms in vSphere attached to the datastores where our Citrix XenDesktop VM's are located. These occur daily at set times on datastores, but the there is a pattern occurring. The datastores are busy from the performance charts I can see but nothing extraordinary. The pattern seems to follow the boot times of the pupil machines (which I can understand). The alarm threshold is set at 75% disk usage, but not for any condition length at present.
Each datastore is 186GB, with base images at the respective 23GB. The VM images are 16MB with each identity disk. VMware sees them as overprovisioned, but a quick calculation sees that this is fine. There are 10-11 VM's per datastore depending on setup.
We have 3 distinct Catalogues and Assignments within Desktop Studio
Pupils - Pooled Random
Staff - Pooled Random
Smartboard - Pooled Random
3 Gold Master Images - related to the 3 assignments, with settings as follows:
There are 35 staff VM's which are running 24/7.
There are 2 smartboard VM's which are also running 24/7
There are 130 pupil VM's which are running in a staggered rotation, with power assignment schedule, leaving 20 running 24/7. This increases as follows:
08:00 upto 60 machines
09:00 upto 90 machines
10:00 upto 130 machines
11:00 upto 130 machines
and vice versa after 17:00.
I have tried the following to remove the problem:
Reducing the number of pupil VM's running from 130 down to 100, to see if its load on the datastores. The alarms still persist
Staggering the machines boot in the power management for the pupil assignment to be:
08:00 upto 40 machines
09:00 upto 60 machines
10:00 upto 90 machines
11:00 upto 100 machines
Still the alarms persist.
Aside from simply adjusting the alarm to solve the alarms popping up
There are currently 1 users browsing this thread. (0 members and 1 guests)