亚洲国产日韩欧美一区二区三区,精品亚洲国产成人av在线,国产99视频精品免视看7,99国产精品久久久久久久成人热,欧美日韩亚洲国产综合乱

Is it hot data? If not, just make a script to run regularly and save the statistical count into another statistical table.
For hot data, see if the index you built before can be turned into a combined index. Because the B-tree starts from the left, if you build an (a, b, c), a and a, b and a, b, and c can be used. Used, the query function at the code layer pays attention to the field order.

Re-optimize the index. There are too many single indexes, so try to build a joint index. If this statistics does not require real-time performance, use md5 (sql statement) as the key, use the result as the value, and store it in the cache (redis).