In Mongoid's QueryCache, if a query with a specified limit is run followed by a query with a smaller limit, it is expected to query again and not use the query cache. However, this behavior is inefficient and QueryCache would be more effective if it could support using a CachingCursor for any queries with smaller limits (currently in Driver, this is supported only when the first query does not specify a limit, in which case following queries with limits will use the cache).