16.5. Configuring a Tiering Volume

When the tier volume is configured to use the cache mode, the configured watermark values and the percentage of the hot tier that is full determine whether a file will be promoted or demoted. The cluster.watermark-low and cluster.watermark-hi volume options set the lower and upper watermark values respectively for a tier volume.

The promotion and demotion of files is determined by how full the hot tier is. Data accumulates on the hot tier until it reaches the low watermark, even if it is not accessed for a period of time. This prevents files from being demoted unnecessarily when there is plenty on free space on the hot tier. When the hot tier is fuller than the lower watermark but less than the high watermark, data is randomly promoted and demoted where the likelihood of promotion decreases as the tier becomes fuller; the opposite holds for demotion. If the hot tier is fuller than the high watermark, promotions stop and demotions happen more frequently in order to free up space.

The following diagram illustrates how cache mode works and the example values you can set.

To set the percentage for promotion and demotion of files, run the following commands:

# gluster volume set VOLNAME cluster.watermark-hi value

# gluster volume set VOLNAME cluster.watermark-low value

You can configure how frequently the files are to be checked for promotion and demotion of files. The check is based on whether the file was accessed or not in the last n seconds. If the promote/demote frequency is not set, then the default value for promote frequency is 120 seconds and demote frequency is 3600 seconds.

To set the frequency for the promotion and demotion of files, run the following command:

# gluster volume set VOLNAME cluster.tier-demote-frequency value_in_seconds

# gluster volume set VOLNAME cluster.tier-promote-frequency value_in_seconds

You can configure the number of reads and writes in a promotion/demotion cycle, that would mark a file HOT for promotion. Any file that has read or write hits less than this value will be considered as COLD and will be demoted. If the read/write access count is not set, then the default count is set to 0.

Set the read and write frequency threshold by executing the following command:

# gluster volume set VOLNAME cluster.write-freq-threshold value

# gluster volume set VOLNAME cluster.read-freq-threshold value

The maximum amount of data that may be migrated in any direction in one promotion/demotion cycle from each node can be configured using the following command:

# gluster volume set VOLNAME cluster.tier-max-mb value_in_mb

If the cluster.tier-max-mb count is not set, then the default data size is set to 4000 MB.

The maximum number of files that may be migrated in any direction in one promotion/demotion cycle from each node can be configured using the following command:

# gluster volume set VOLNAME cluster.tier-max-files count

If the cluster.tier-max-files count is not set, then the default count is set to 10000.