Overview of mmpmon

The mmpmon facility allows the system administrator to collect I/O statistics from the point of view of GPFS™ servicing application I/O requests.

The collected data can be used for many purposes, including:
  • Tracking I/O demand over longer periods of time - weeks or months.
  • Recording I/O patterns over time (when peak usage occurs, and so forth).
  • Determining if some nodes service more application demand than others.
  • Monitoring the I/O patterns of a single application which is spread across multiple nodes.
  • Recording application I/O request service times.
Figure 1 shows the software layers in a typical system with GPFS. mmpmon is built into GPFS.
Figure 1. Node running mmpmon
This is a node with GPFS running, mmpmon. It consists of four layers. The top layer is the application. The second layer is the operating system, either AIX or Linux. The third layer is GPFS with mmpmon as a subset of GPFS. The fourth layer is the NSD layer.
Parent topic: Monitoring GPFS I/O performance with the mmpmon command
Related concepts:
Understanding the node list facility
Understanding the request histogram facility
Understanding the Remote Procedure Call (RPC) facility
Related tasks:
Specifying input to the mmpmon command
Display I/O statistics per mounted file system
Display I/O statistics for the entire node
Reset statistics to zero
Displaying mmpmon version
Related reference:
Example mmpmon scenarios and how to analyze and interpret their results
Other information about mmpmon output