CPFS Performance Monitoring & Tuning Overview - Cloud Parallel File Storage

How do I view the performance metrics of a CPFS file system?

Log on to the NAS console. On the Performance Monitoring page, select CPFS from the File System Type drop-down list to view metrics such as throughput, IOPS, and latency for read and write operations, and metadata QPS. For more information, see View CPFS performance monitoring.

How do I optimize the performance of a CPFS file system?

Performance parameters

Parameter	Description
workerThreads	The number of client threads. Increase the value of this parameter to improve the IOPS.
maxStatCache	The maximum number of cached file properties. Optimize the parameter setting to improve the performance of the CPFS file system when you need to frequently read file properties.
maxFileToCache	The maximum number of cached inodes. Optimize the parameter setting to improve the performance of the CPFS file system when you need to access many small files multiple times.
prefetchPct, prefetch, and write-behind buffer	The maximum percentage of the page pool that can be used. Optimize the parameter setting to improve the performance of the CPFS file system in the sequential read/write scenarios.
pagepool	The size of page pool memory that the client reserves. Optimize the parameter setting to leave a larger storage space for cached data. The minimum value of this parameter is 4 GB.
maxMBpS	The maximum throughput of the client.

Configuration steps
Command format
```
/usr/lpp/mmfs/bin/mmchconfig <parm>=<value> -N <node-list>
```
1. The following example shows how to configure workerThreads.
```
/usr/lpp/mmfs/bin/mmchconfig workerThreads=1024 -N ecs1
```
2. Run the following commands on the modified node to restart the CPFS client. The new configuration takes effect after the client restarts.
  Important
  The restart operation remounts the CPFS file system on the node and interrupts I/O services.
  1. Stop
```
/usr/lpp/mmfs/bin/mmshutdown -N <node-list>
```
  2. Start
```
/usr/lpp/mmfs/bin/mmstartup -N <node-list>
```
Tuning recommendations
We recommend that you optimize the parameter settings based on the configurations of the compute nodes and your business requirements on I/O. Increasing the values of some parameters such as cache-related parameters can make the CPFS file system use more storage space.