Cache usage

GPFS creates a number of cache segments on each node in the cluster. The amount of cache is controlled by three attributes.

These attributes have default values at cluster creation time and might be changed through the mmchconfig command:

pagepool

The GPFS pagepool attribute is used to cache user data and file system metadata. The pagepool mechanism allows GPFS to implement read and write requests asynchronously. Increasing the size of the pagepool attribute increases the amount of data or metadata that GPFS can cache without requiring synchronous I/O. The operating system and other software that is running on the node might restrict the amount of memory available for GPFS on a particular node.

The optimal size of the pagepool attribute depends on the needs of the application and effective caching of its reaccessed data. For systems where applications access large files, reuse data, benefit from GPFS prefetching of data, or have a random I/O pattern, increasing the value for the pagepool attribute might prove beneficial. However, if the value is set too large, GPFS starts with the maximum that the system allows. See the GPFS log for the value it is running at.

To change the size of the pagepool attribute to 4 GB:

 mmchconfig pagepool=4G

maxFilesToCache

The total number of different files that can be cached at one time. Every entry in the file cache requires some pageable memory to hold the content of the file's inode plus control data structures. This is in addition to any of the file's data and indirect blocks that might be cached in the page pool.

While the total amount of memory that is required for inodes, attributes and control data structures varies based on the functions that are being used, it can be estimated as a maximum of 10 KB per file that is cached.

Valid values of maxFilesToCache range from 1 through 100,000,000. For systems where the applications use many files, of any size, increasing the value for maxFilesToCache might prove beneficial. This is true for systems where many small files are accessed. The value must be large enough to handle the number of concurrently open files plus allow caching of recently used files.

If the user does not specify a value for maxFilesToCache, the default value is 4000.

maxStatCache

This parameter sets aside extra pageable memory to cache attributes of files that are not currently in the regular file cache. This is useful to improve the performance of both the system and GPFS stat() calls for applications with a working set that does not fit in the regular file cache. For systems where applications test the existence of files, or the properties of files without opening them, as backup applications do, increasing the value for maxStatCache can improve performance.

The memory that is occupied by the stat cache can be calculated as:

maxStatCache × 480 bytes

The valid range for maxStatCache is 0 - 100,000,000. If you do not specify values for maxFilesToCache and maxStatCache, the default value of maxFilesToCache is 4000 and the default value of maxStatCache is 1000. If you specify a value for maxFilesToCache but not for maxStatCache, the default value of maxStatCache is 4 * maxFilesToCache or 10000, whichever is smaller.

In versions of IBM Spectrum Scale earlier than 5.0.2, the stat cache is not effective on the Linux® operating system unless the Local Read-Only Cache (LROC) is configured. For more information, see the description of the maxStatCache parameter in the topic mmchconfig command.

The total amount of memory GPFS uses to cache file data and metadata is arrived at by adding pagepool to the amount of memory that is required to hold inodes and control data structures (maxFilesToCache × 10 KB), and the memory for the stat cache (maxStatCache × 480 bytes) together. The combined amount of memory to hold inodes, control data structures, and the stat cache is limited to 50% of the physical memory on a node that is running GPFS.

During configuration, you can specify the maxFilesToCache, maxStatCache, and pagepool attributes that control how much cache is dedicated to GPFS. These values can be changed later, so experiment with larger values to find the optimum cache size that improves GPFS performance without negatively affecting other applications.

The mmchconfig command can be used to change the values of maxFilesToCache, maxStatCache, and pagepool. The pagepool parameter is the only one of these parameters that might be changed while the GPFS daemon is running. A change to the pagepool attribute occurs immediately when you are using the -i option on the mmchconfig command. Changes to the other values are effective only after the daemon is restarted.

For more information on these cache settings for GPFS, see GPFS and memory.