Administrator Guide

Performance monitoring using Unisphere for metro node
The performance monitoring dashboard provides a customized view into the performance of your system. You decide which
aspects of the system's performance to view and compare.
Figure 9. Performance monitoring dashboard (for HTML5)
Performance information for the current 5-minute window is displayed as a set of charts, including:
● WAN Link Performance chart - Shows the WAN link performance for the cluster to you are connected to. Use this chart
to monitor link performance to help determine the bandwidth requirements for your specific environment, gather statistical
data over time, monitor network traffic during peak periods, or to plan data mobility jobs to avoid peak usage times.
● WAN Latency chart - Provides a time-based view of the WAN Latency. The categories avg-lat/min-lat/max-lat each report
values observed in the last 5 seconds or less.
● Write Latency Delta chart - Provides the delta between Front-end latency and Back-end Latency per director. This is a
key metric for Local/Metro – the amount of overhead time metro node spends processing a write.
● Back-end Errors chart - Displays the back-end I/O errors to and from the storage array. There are three categories of
back-end errors: Aborts, timeouts, and resets.
● Back-end Throughput chart - Shows the back-end I/Os per second over time for directors. Generally throughput (or more
commonly referred to as IOPS) is associated with small block I/O (4KB or 16KB I/O requests.)
● Back-End Bandwidth chart - Shows the quantity of back-end reads and writes per second over time for directors.
Generally bandwidth (measured in KB/s or MB/s) is associated with large block I/O (64KB or greater I/O requests).
● Back-end Latency chart - Provides details of the back-end latency statistics for your metro node system in graphical
form over time. The chart allows you to view current or historical performance data that you can use to monitor peaks in
workload, detect performance issues, or view what was happening in the system when a specific problem occurred.
● Rebuild Status dashboard - Display the status of any rebuilds or migration operations that are running on your metro node
system.
● CPU Utilization chart - Provides a time-based view of the utilization load on the primary director CPU on your metro node
system. By default, the chart shows an averaged view of the utilization loads of all the directors in your metro node system.
● Heap Usage chart - Shows a percentage of the heap memory used by the firmware on a director.
● Front-end Aborts chart - Displays the number of aborts per second over time for directors on your metro node system. By
default, the chart shows averaged front-end aborts for the metro node system.
● Front-End Bandwidth chart - Displays the quantity of front-end reads and writes per second over time for directors on
your metro node system. By default, the chart shows the total front-end bandwidth for the metro node system.
● Front-end Latency chart - Provides details of the front-end latency statistics for your metro node system in graphical
form over time. The chart allows you to view current or historical performance data that you can use to monitor peaks in
workload, detect performance issues, or view what was happening in the system when a specific problem occurred.
● Front-end Queue Depth chart - Provides the count of front-end operations per director. It describes the number of
concurrent outstanding operations active in the system.
● Front-End Throughput chart - Displays the front-end I/Os per second over time for directors on your metro node system.
By default, the chart shows the total front-end throughput for the metro node system.
● Virtual Volume Throughput chart - Provides a time-based view of the total throughput or IOPS for a virtual volume.
Generally throughput, more commonly referred to IOPS, is associated with small block I/O (512B to 16KB I/O requests.
● Virtual Volume Latency chart - Provides a time-based view of the IO Latency for a virtual volume broken down by read
and write latency. Virtual volume latency is defined as the amount of time an I/O spends within metro node for a given
virtual volume.
Performance and Monitoring
83