Request histogram (rhist) output - how to aggregate and analyze the results

The rhist requests are used to produce histogram data about I/O sizes and latency times for I/O requests.

The output from the rhist requests can be used to determine:
  1. The number of I/O requests in a given size range. The sizes may vary based on operating system, explicit application buffering, and other considerations. This information can be used to help determine how well an application or set of applications is buffering its I/O. For example, if there are many very small or many very large I/O transactions. A large number of overly small or overly large I/O requests may not perform as well as an equivalent number of requests whose size is tuned to the file system or operating system parameters.
  2. The number of I/O requests in a size range that have a given latency time. Many factors can affect the latency time, including but not limited to: system load, interconnection fabric load, file system block size, disk block size, disk hardware characteristics, and the operating system on which the I/O request is issued.

For information on interpreting mmpmon output results, see Other information about mmpmon output.