Dell EMC PowerVault MD 34XX/38XX Series Storage Arrays Administrator's Guide

You can collect the following performance data:

• Total I/Os – Total I/Os performed by this object since the beginning of the polling session.
• I/Os per second – The number of I/O requests serviced per second during the current polling interval—also called an I/O request rate.
• MBs per second – The transfer rate during the current polling interval. The transfer rate is the amount of data in megabytes that can be moved through the I/O data connection in a second—also called throughput.   
  NOTE: A kilobyte is equal to 1024 bytes, and a megabyte is equal to 1024 x 1024 bytes. Some applications calculate kilobytes as 1,000 bytes and megabytes as 1,000,000 bytes. The numbers reported by the monitor might be lower by this difference.
• I/O Latency – The time it takes for an I/O request to complete, in milliseconds. For physical disks, I/O latency includes seek, rotation, and transfer time.
• Cache Hit Percentage – The percentage of total I/Os that are processed with data from the cache rather than requiring I/O from disk. Includes read requests that find all the data in the cache and write requests that cause an overwrite of cache data before it has been committed to disk.
• SSD Cache Hit Percentage – The percentage of read I/Os that are processed with data from the SSD physical disks.

The metrics available include the current value, minimum value, maximum value, and average value. The current value is the most recent data point collected. The minimum, maximum, and average values are determined based on the start of performance monitoring. For real-time performance monitoring, the start is when the Array Management Window (AMW) opened. For background performance monitoring, the start is when background performance monitoring started.

Performance metrics at the storage array level are the sum of metrics on the RAID controller modules. Metrics for the RAID controller module and disk group are computed by aggregating the data retrieved for each virtual disk at the disk group/owning RAID controller module level. The values reported for a RAID controller module, or a storage array might be greater than the sum of the values reported for all the virtual disks. The values reported for a RAID controller module or storage array include both host I/Os and I/Os internal to the storage array (metadata reads and writes), whereas the values reported for a virtual disk include only host I/Os.

On a performance monitor graph, you can specify one metric and up to five objects. Not all metrics apply to all objects.