When hypervisor monitoring is enabled for VMware or XenServer, all of the displays in the technology will show resource utilization in regards to the hypervisor - not Windows. For Example, the Dashboard>CPU/Memory/Storage pages, Snapshot drill down metrics, View>XenApp & XenDesktop, etc.
When hypervisor monitoring is enabled and you are reviewing these pages to identify issues in the environment, it is suggested that you focus in on the “average” metric and not the “current” data point that Goliath is getting back from the hypervisor.
For example, you could see 99-100% CPU utilization from Windows, but the Dashboard>CPU display was only showing 5.22% for current CPU (see below screenshots)
Windows Task Manager:
Dashboard>CPU:
This discrepancy could cause concern in that the machine displays 99% utilization and Goliath is only 5.22%. But, even though Goliath is showing the current as 5.22%, it is important to look at the average, being 42%, which indicates that there are issues taking place and the utilization is increasing.
When reviewing the performance metrics, it is suggested to not compare VMware metrics to the Windows metrics because they are collected based on different mechanisms and intervals so the graphs for the same time period are not necessarily showing the same data and information.
For instance VMware metrics are collected agentless using our API integration on an interval of every 5 mins. Windows metrics are collected by the agent every 60 to 120 seconds based on how your environment is configured. Also, the Dashboard>CPU/Memory/Storage pages automatically refresh every 5 mins.
When on the different pages of the technology, you can click on the chart icon in the 'gph' column to drill down into the snapshot metrics. By looking at the machine and its past 6 hours of performance, one can instantly see that there are issues taking place with CPU.
To avoid any confusion, it is always recommended to enable email/text alert notifications so that you are made aware of an issue immediately and do not by have to keep an eye on the dashboards. If Windows metrics are most important to you, configuring a "CounterWatch" type alert reports on Windows Perfmon counters and allows you to set sustained durations to make the alerts actionable. The technology also has alerting for VMware metrics, but as mentioned previously this information is not a granular.