Skip to main content

vHost performance summary

Home > Select Project > Virtual environment > vHost performance summary

When you run several ESXi hosts, it is hard to tell at a glance which host the load is concentrated on. vHost performance summary places one ESXi host per row and compares CPU, memory, disk, and network metrics in the same table. The layout assumes a flow where you compare values to shortlist hosts that may be a problem, then move on to vHost performance details to narrow down the cause. It is intended for operators who review resource allocation in virtualization-based infrastructure.

Note

For data to appear

The data on this screen is filled only after you register vCenter or ESXi Host connection information in VM collection settings. If the table is empty, the screen also shows a link to the collection settings.

Table layout

The table columns are bundled into five groups and displayed with a two-level header.

GroupDescription
vHost informationInventory information such as host name, vCenter address, datacenter, cluster, and folder. Fixed on the left
CPUCPU metrics such as usage, ready time, latency, demand share, and Co-Stop
MemoryMemory metrics such as memory usage, active memory usage, and balloon (vmmemctl)
DiskDisk usage and maximum total latency
NetworkNetwork usage and send/receive metrics

The host name column in the vHost information group cannot be unchecked. Without the name, you could not tell what the table is a list of.

Selecting columns

Select the metrics to display per group in the Column settings menu. Hover over each metric to see a description of what it means. Several pairs are hard to tell apart by name alone (for example, the ready time series and the latency series), so we recommend reading the descriptions as well.

Refresh interval and query time

This screen reloads the latest values every 30 seconds. The agent's collection interval is 20 seconds, so asking more frequently would only repeat the same values.

If you pause the real-time query, you can select a past point in time. You can go back as far as the raw data retention period of the project, and dates beyond the retention period cannot be selected in the calendar. The information icon next to the picker shows how many days are available.

Caution

The retention period differs by project

The raw data retention period depends on the project settings. Use the number of days shown on the screen as the reference. The retention period of aggregated data (5-minute and 1-hour units) is much longer, but this screen queries raw data, so the raw data criterion applies.

When a value is shown as a dash

The memory usage cell may appear as -. When the agent cannot obtain the total physical memory, it sends the usage as 0, and that 0 means "unknown" rather than "0%". The screen shows a dash instead of drawing it as 0%. This keeps a host that barely uses memory from being confused with a host whose value could not be read.