Live! Cluster Utilization Dashboard
The
Live! Cluster Utilization
dashboard complements the
Cluster Performance
dashboard. Use this dashboard to view the
clusters that are working excessively and are close to their physical limit. This dashboard
displays ESXi hosts that have CPU or memory saturation that can lead to performance issues
for the VMs running on the host.Design Considerations
This dashboard is designed to complement
the
Live! Cluster Performance
dashboard and it shares design
considerations.How to Use the Dashboard
As this dashboard has an identical design with the
Live! Cluster
Performance
dashboard, it has the same usage procedure. Unlike the
heat maps in the Live! Cluster Performance
dashboard, the three
heat maps in this dashboard have a different scale, reflecting the different nature
of the counters.Logically, memory is a form of storage.
It acts as a cache to disk as it is much faster. A high utilization is better, as it
indicates that more data is being cached. The ideal situation is when ESXi host
Consumed metric is red but ESXi host Ballooned metric is green. When Ballooned is
red and Consumed is gray, it means that there was high pressure in the past but it
is not there anymore. The reason the ballooned stays red is because the ballooned
pages were never requested back.
The ballooned memory counter was selected
over the swapped or compressed memory counters as it is a better leading indicator.
Since all three can co-exist at the same time, they are displayed in the line chart.
Ballooned is displayed in absolute amount and not as a percentage, because the
higher the size, higher are the chances for it to impact a VM. If you feel using
percentage is easier for your operations, create a super metric to translate the
value.
The heat map displays Wastage by a new
color. The dark gray color indicates that wastage as capacity is not used. The
performance problem due to low utilization can be caused by a bottleneck
elsewhere.
Analyze if the ESXi host is contributing.
A light gray box indicates that the host is a part of the cluster but there is no
utilization. It is possible for the host to be in the maintenance mode or is powered
off.
Points to Note
- ESXi host chooses to swap over compression if the compression ratio is less than 4x.
- If the ESXi host's physical NIC is saturated in your environment, then you can add a Network Throughput heat map.