The numbers reported by vmstat inside a Linux guest report on the virtual CPU usage. Unless you have attached real CPUs to a guest, they're not a good representation of real CPU utilization -- it tells you only what the division of labor is from Linux's perspective. Useful for determining whether you have an I/O hog or not, but not reflective of the actual usage of the system.
If you are using Perfkit, you need to install RMFPM in the guest and configure Perfkit to accept data from it to get a better sense of what is actually going on. There are other solutions that use a different approach to correlate the VM and Linux CPU numbers (SNMP + VM accounting records), and recent mods to the Linux kernel (I think those made it into SLES 9 SP4, but don't know for certain -- they're in SLES 10 SP1) that provide more accurate VM accounting records, but I don't know if Perfkit supports them yet. Don't run RMFPM for very long; it's a real hog on it's own. ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390