Cache Learning Mode

The data cached in the core flows is held in ESQL ROW variables. These variables are integration bus tree data types and are very similar to doubly linked lists. The performance of accessing data in the caches is dependent on the order of data in the cache and how often you look for it. Ideally, the data you refer to most frequently would be the data nearest to the top of the cache.

The APP_VERSION table has a column called CACHE_LEARN, which, if set to Y, tells the caches to count which items in the cache are accessed most frequently. When the FlowManager polls the APP_VERSION table, if cache learning is enabled and the cache does not need to be reloaded, the cache will instead be reordered based on the current cache access counts. This continues until cache learning is again disabled using the APP_VERSION table.

Cache learning mode can be used to optimize the storage of cached data based on real time transactions for a short time and can then be disabled once it has 'learned'.

Currently, it is not possible to persist or predefine cache access statistics.