When it cannot flatten or materialize a subquery, Adaptive Server uses an in-memory cache to store the results of each evaluation of the subquery. While the query runs, Adaptive Server tracks the number of times a needed subquery result is found in cache. This is called a cache hit ratio. If the cache hit ratio is high, it means that the cache is reducing the number of times that the subquery executes. If the cache hit ratio is low, the cache is not useful, and it is reduced in size as the query runs.
Caching the subquery results improves performance when there are duplicate values in the join columns or the correlation columns. It is even more effective when the values are ordered, as in a query that uses an index. Caching does not help performance when there are no duplicate correlation values.