Seeking details on RESULT_CACHE statistics

Question

I'm running 11.2.0.1, and I'm just beginning to look into the server result caching feature introduced in 11g. I'm evaluating the basic effectiveness of queries and the cache by watching query response times and monitoring v$result_cache_statistics. So far, I've found that queries with aggregation work well with the cache, and "SELECT *" queries from small tables also use the cache. Unfortunately, the type of query I really need to work so I can gain the benefit of the speed caching offers often ends up increasing the "Delete Count Invalid" statistic, without providing any benefit. My first assumption was that the cache itself was too small for the query results. My test query uses a table that is roughly 450MB. However, even after increasing RESULT_CACHE_MAX_SIZE to 500MB, the query results were still not cached.
I've read what I can find in the documentation and Metalink about the result cache feature and v$result_cache_statistics, but I haven't found any discussion of what causes some queries to be cached and used (where Find Count increases and performance is obviously improved) and others to not be used (where Delete Count Invalid increases and the query doesn't benefit). Is there another source of information that would explain these statistics and what they mean? The [reference page for the dynamic performance view|http://download.oracle.com/docs/cd/E11882_01/server.112/e10820/dynviews_2140.htm#sthref3336] doesn't provide the detail I'm looking for.
Thanks -
Wally

Answer 1

Ah! RESULT_CACHE_MAX_RESULT was the key, as far as I can tell. By stretching this out (probably further than it should be stretched, I admit), I am able to see the query results in the cache statistics where I want them, with the corresponding increase in performance.
I acknowledge the previous contributor who questioned my logic in caching such a large portion of the data. At this point, I'm just working to understand the technology and its limitations so I can apply it rationally. To this end, I'm still seeking some documentation that can explain the specific statistics in V$RESULT_CACHE_STATISTICS in greater detail than the Oracle Database Reference http://download.oracle.com/docs/cd/E11882_01/server.112/e10820/dynviews_2140.htm book offers.
SQL> select * from v$result_cache_statistics;
        ID NAME                                               VALUE
         1 Block Size (Bytes)                                 1024
         2 Block Count Maximum                                512000
         3 Block Count Current                                512000
         4 Result Size Maximum (Blocks)                       384000
         5 Create Count Success                               8
         6 Create Count Failure                               0
         7 Find Count                                         6
         8 Invalidation Count                                 0
         9 Delete Count Invalid                               4
        10 Delete Count Valid                                 2
        11 Hash Chain Length                                  1
11 rows selected.I'm currently most interested in numbers 8 through 11, what they represent, what query conditions increment them, and so forth. Any pointers to documentation or Metalink notes along these lines will be greatly appreciated.
Thanks!
Added link to documentation: user650205 on Feb 10, 2010 9:44 AM