ANSWER:

Observations

• Your question says you have 426870 for Qcache_total_blocks. Since each block is 1K, blocks are aligned to the nearest multiple of 1024. Thus, your query cache has about 417M.
• If you divide Qcache_not_cached (7128902) by Qcache_inserts (49557287), this shows that about 14.4% (1 out of 7) of your queries could not be cached.
• (Qcache_total_blocks (426870) – Qcache_free_blocks (35327)) / Qcache_queries_in_cache (195659) = 2K
• On average, each query cache result hold about 2K, which is theoretically not possible. This could indicate some fragmentation in the query cache.

Recommendations

• You need a bigger value for query_cache_size (perhaps 2G). If you are using InnoDB, you may get some performance increase if you disable the query cache altogether by setting query_cache_size to 0.
• You need a smaller value for query_cache_min_res_unit (minimum value is 512 (0.5K)) if you want to hold small result sets.
• You need a bigger value for query_cache_limit if you want to cap how big a result is too big.
• See my earlier post : Is the overhead of frequent query cache invalidation ever worth it?

RolandoMySQLDBA from http://dba.stackexchange.com/questions/43003