Let me preface this post by saying I understand there have been a lot of questions, articles and white papers written about CPU Ready times in ESX but I have yet to find one that helps me understand what I am seeing on my own cluster.
Here is my scenario... I have a 2-vCPU Windows Terminal Server running on a 4-host ESX cluster. Each server in the cluster has a single QC 2.1GHz CPU. The cluster has a total of 19 VMs, none of which are particularly CPU-intensive. CPU usage in the cluster is low enough that DPM keeps two of these hosts powered down at all times.
The Terminal Server, which routinely has between 6 and 12 people logged on, shows CPU Ready times of thousands and sometimes tens of thousands of milliseconds (i.e. 20,000ms) of CPU Ready times. Based on my understanding of CPU Ready this means that the vCPUs in the VM are waiting several to many +seconds +before being able to execute the next instruction. If this were the case, doesn't that mean the VM would basically be frozen until the next instruction could be executed? I would think that would make something like a Terminal Server completely unusable to those who are logged in yet I never hear any complaints. I have attached CPU graphs from both the host and VM.
Since there is no way users would wait literally seconds before their next action I can only assume my understanding of CPU Ready times is incorrect but I cannot find any other description. Could somebody help me understand this better? Thanks.