1 Reply Latest reply on Dec 4, 2007 2:18 PM by virtualdud3

    ESXTOP and %RDY - Making sense of the statistics

    CiscoKid Hot Shot

      All, other than multiple vCPUs on a given VM causing excessive wait times for the processing of instructions, what else can cause high values of %RDY?  I have read that anything about 5.00 is considered to be reason enough to begin investigating the suspect VM is this true? Is there something internal to the VM that could be causing high value %RDY?  At what duration should an administrator be alarmed?  Lastly, could a VM running at higher than 5% cause others to become slow performing?  Thanks in advance.

        • 1. Re: ESXTOP and %RDY - Making sense of the statistics
          virtualdud3 Master

          Well, at the most-basic level high CPU Ready values can be caused by simply running multiple VMs on the same ESX server, and the ESX server is "running out" of CPU cycles to allocate to the VMs.  Or, if you are using a VM running applications that require a great deal of processing, this can also cause the VM to have a high CPU ready value.

           

          I am not sure there are really any hard and fast rules regarding CPU ready value.  The value of 5% was from a VMware pdf, so it is something to consider, but that isn't really an absolute value.  If you have a server that is rarely used, and users won't notice if the server is being heavily loaded, then 5% CPU ready may not be a problem.  If the VM is running an often-used, critical application, then users may notice if the VM is "waiting" for a processor 5% of the time it is running.  Like a lot of things, "it depends".

           

          I'm not sure that a VM running at higher than 5% CPU could cause other VMs to be slow-performing.  However, when a VM is running at a high value of CPU ready, this is an indication that the ESX host is lacking CPU resources, and the other VMs will likely encountering the same issue.

           

          If only a single VM is consistently experiencing high values of CPU ready, you may want to consider either granting the VM a CPU reservation and/or increasing its CPU shares (either at the VM level or via the use of resource pools).

           

          As an aside, if you wish to monitor CPU ready via the VI client, you might wish to increase the level of "Statistics Logging" from the default value of "2" to "4", as this will allow you to monitor CPU ready for time intervals other than real-time.  And, when viewing CPU ready via the VI client, the value of 5% will correspond to a value of 1000ms (the graph refresh interval is (by default) 20 seconds, 5% of 20 seconds is 1 second, and 1 second is 1000 milliseconds).

           



          ###############

          If you found this information useful, please consider awarding points for "Correct" or "Helpful".  Thanks!!!