Performance monitors collect and displays statistics to determine how a port or volume is being used, how much I/O is being processed, CPU usage, and so on.
Performance monitoring is supported in both the VPlexcli and Unisphere, and falls into three general types:
Current load monitoring allows administrators to watch CPU load during upgrades, I/O load across the inter-cluster WAN link, and front-end compared to back-end load during data mining or back up.
Current load monitoring is supported in Unisphere.
Long term load monitoring collects data for capacity planning and load balancing.
Long term load monitoring is supported by monitors created in the CLI and/or perpetual monitors.
Troubleshooting monitoring helps identify bottlenecks and resource hogs.
Troubleshooting monitors are supported by monitors created in the CLI and/or perpetual monitors.
NOTE:In Unisphere for metro node, performance statistics are displayed per cluster. To view statistics for both clusters in a Metro configuration, connect to both clusters.
Custom monitors
You can use the CLI to create custom monitors to collect and display selected statistics for selected targets.
GeoSynchrony includes perpetual monitors that gather a standard set of performance statistics every 30 seconds. Perpetual monitors collect the statistics related to the performance of metro node directors and virtual volumes.
Perpetual monitor files are collected as part of
collect-diagnostics. Collect-diagnostics is per cluster, so in Metro configurations, run the command from one node on each metro node cluster.
Output of perpetual monitors is captured in the file
smsDump_date.zip inside the base collect-diagnostics zip file.
Within
smsDump_date.zip file, monitor files are in
clilogs/.
You can also copy the perpetual files from the management server. They are located in
/var/log/VPlex/cli/. There is one perpetual monitor file per director, identifiable by the keyword “PERPETUAL” .
The following is an example for the statistics that perpetual monitors collect on virtual volumes: