Doris's current statistics are only inaccurate table rows and table size. Due to the lack of some more accurate column statistics, the Doris query plan cannot evaluate and select a better plan. Also the current statistical information is not a complete structure.
I propose to add a statistical information collection module in the query layer. Mainly responsible for the collection, storage, use and other functions of statistical information, so as to assist the optimizer to obtain better query planning. Ling Miao