ESXi “esxtop” data can be huge, especially if you capture all columns. Not all metrics are equally important. Typically, for diagnosing performance in a VM environment, you want to focus on:
-
CPU Usage (e.g.,
%USED,%RDY,%CSTP/ co-stop for SMP VMs, etc.)%RDY(ready time) is a key metric to see if the VM is waiting for CPU.%CSTP(co-stop) indicates scheduling delays for multi-vCPU VMs.
-
Memory Usage (e.g.,
MEMCTL,MEM_ACTIVE,MEM_CONSUMED,MEM_OVERHEAD)- If memory is overcommitted, ballooning or swapping might occur.
-
Network (if relevant)
- Bytes transmitted/received if you suspect network bottlenecks for the map application.
-
Disk / Latency
- Important if the application is I/O sensitive.
- CPU:
%USED,%RDY,%CSTP,%WAIT - MEM:
%MEM,MEMCTL,SWAP,BALLON - NET metrics if the map data is large and network-latency-sensitive.
Consult VMware docs or your internal knowledge base for the recommended columns that best reflect CPU scheduling issues or memory pressure. Typically, CPU Ready Time (%RDY) is one of the most critical. High ready times often indicate CPU contention on the host.