It has long been discussed when to go for a virtual desktop solution based on 1 or 2 vCPUs vCPUs. Well, it's not a rule, but I hope it can help shed some light for others.
I recently during a stress test for a client where the objectives were diversed, but I would like to highlight the result of running the same workloads with 1 vCPU and 2, with the same amount memory. . And how it greatly influenced the behavior of the boot storm
The test scenario was based on:
- 32 hosts Cisco UCS running VMware vSphere 4.1
- 00 virtual desktops provisioned with Provisioning services 5.6 SP1
- vDisk stores were CIFS based on NetApp boxes
- VM were :, 1 GB of RAM and 6GB PVS write caching on the base Windows XP SP3 virtual disk
with 1 vCPUs vitual based workstations, the entire infrastructure could, but in about 45 minutes , which under certain circunstances we were able to have after a storm starting around 00 unregistered virtual desktops when the infrastructure has reached its equilibrium state. Even refine the value of the Timeout startup services on virtual desktops.
With 2 vCPUs virtual desktops based, it started in 25 minutes , with 12 unregistered virtual desktops at the end cycling process
this result, I can say definitely, additional CPU will :.
- stimulate the diffusion / start process
- improve responsiveness and registration of virtual desktops
- with an increase in the cost of CPU hit the hypervisior
therefore, delivery will be:
- If you plan to go big with a huge number of virtual machines lifecycling every day ...
- If you have a large number of workstations that you might need to provide in advance ...
- If the window of your bike must be the shortest possible ...
... in any case, additional CPU will improve your cycling process, to reduce the gap virtual recording desktops with an additional cost of higher peak use of the host processor, improving your infrastructure availability.
0 Komentar