Use the esxtop command-line utility to periodically monitor the CPU and memory usage of individual hosts and their VMs. The esxtop utility provides a detailed look at how ESX uses resources in real time.
Here are some examples and guidelines for interpreting host-related esxtop data:
If the load average on the first line of the esxtop CPU panel is equal to the number of physical processors in the system, the host is overloaded.
If the usage percentage for the physical CPUs on the PCPU line (PCPU USED(%)) is 90% or greater, the CPUs are approaching an overloaded condition.
Here are some examples and guidelines for interpreting VM-related esxtop data:
High values for ready wait (%RDY) and %CSTP indicate contention for CPU. %RDY should be less than 10%-15% per vCPU. %CSTP should be less than 5% per vCPU.
VMs with a high %CSTP might indicate that they have more vCPUs than required. For an IP Availability ManagerVM with four vCPUs and a high %CSTP, consider reducing the vCPUs to two.
Also, check the esxtop data for memory ballooning and swapping. (In the CPU panel, check the swap wait %SWPWT value; in the Memory panel, check the MCTLTGT (MB) value, the SWCUR (MB) value, and other balloon- and swap-related lines.) Swapping can be avoided in a VM by reserving memory for the VM that is at least equal in size to the VM’s active working set.