摘要 |
Providing cost model data for tuning query cache memory size in database systems. In one aspect, a query is received, and a query cache is checked for a stored query matching the received query. In response to a cache miss occurring, a token representing the received query is generated and checked for matching a token stored in a simulation area, each stored token representing a query object previously evicted from the query cache. A processing time saved by an avoidance of the cache miss is determined, and a cache size needed to avoid the cache miss is determined only when the generated token matches a stored token, by summing sizes of all query objects represented by the stored tokens. Each stored token is included in a stub that also includes the size of the represented query object and an identification number indicating a stored order.
|