Autocaching in Data Virtualization

Autocaching improves query performance by automating the entire cache lifecycle from creating to deleting caches.

Autocaching improves the performance of your queries by automatically caching the results of frequently used queries, based on the cache recommendations that the cache recommendation engine provides. Autocaching also evicts old caches if they are no longer beneficial.

With this feature, you can customize the following settings:
  • The name of each autogenerated cache.
  • How often autocaching runs.
  • The amount of storage space that caches can occupy.
  • The type of queries in your workload that you want autocaching to analyze.

Autocaching is disabled by default and you can enable it. For more information, see Enabling autocaching in Data Virtualization.

Creation and eviction model

Autocaching creates the top ten caches that the cache recommendation engine recommends, ensuring that the most critical caches are created first. If the cache recommendation engine provides more than ten recommendations, the remaining caches are created in the next run, as long as the recommendations remain the same and in the same order. If in the next run, the recommendation engine changes its recommendations or the order of the caches, autocaching will follow the new list and create the first ten caches from that list instead. Also, autocaching does not create duplicate caches.

For example: If the cache recommendation engine recommends 15 caches to create, autocaching creates the first ten in the initial run. If in the next run, the recommendation engine recommends the same 15 caches, then autocaching creates the remaining five caches.

Autocaching only evicts autogenerated caches and does not delete user-defined caches, however users can always delete or disable autogenerated caches. Autocaching might evict autogenerated caches based on the following criteria:
  • Autogenerated caches that failed to be created.
  • Autogenerated caches that have not been used for at least 15 days.
  • Autocaching does not have enough storage available, which is defined by the storage limit parameter in the autocaching settings. In this case, autocaching first evicts caches that the cache recommendation engine no longer recommends. Afterwards, autocaching evicts caches starting from the lowest-ranked to the higher-ranked ones until there is enough storage space.

To learn more about the cache recommendation engine, see Cache recommendations in Data Virtualization.