Determining sectors of a track to stage into cache by training a machine learning module

A system for determining tracks to stage from a storage, comprising:

• a processor;
• a cache;
• a computer readable storage medium having computer readable program code that when executed by the processor causes operations, the operations comprising:
• providing a machine learning module that receives as input performance attributes of system components affected by staging tracks from the storage to the cache and outputs a staging strategy comprising one of a plurality of staging strategies, wherein each staging strategy indicates at least one of a plurality of sectors of a track to stage into the cache;
• determining a margin of error based on a current value of a performance attribute and a threshold of the performance attribute;
• determining an adjusted staging strategy of the plurality of staging strategies based on the margin of error;
• retraining the machine learning module with current performance attributes to output the adjusted staging strategy;
• and
• using the retrained machine learning module to output one of the staging strategies to use to determine sectors of a track to stage into the cache for a requested track not in the cache.

A method for determining tracks to stage from a storage, comprising:

• providing a machine learning module that receives as input performance attributes of system components affected by staging tracks from the storage to a cache and outputs a staging strategy comprising one of a plurality of staging strategies, wherein each staging strategy indicates at least one of a plurality of sectors of a track to stage into the cache;
• determining a margin of error based on a current value of a performance attribute and a threshold of the performance attribute;
• determining an adjusted staging strategy of the plurality of staging strategies based on the margin of error;
• retraining the machine learning module with current performance attributes to output the adjusted staging strategy;
• and
• using the retrained machine learning module to output one of the staging strategies to use to determine sectors of a track to stage into the cache for a requested track not in the cache.