What is a federated index?

It refers to indexing multiple columns on a table , The union index is also a B+ Trees , The difference is that the number of key values in the union index is not 1, But greater than or equal to 2.

Left most matching principle
Assume that the joint index of the above figure is （a,b）. The union index is also a B+ Trees , The difference is B+ The tree is indexing a Sort on the basis of , On the index b Sort . So the data follows （1,1),(1,2)… Sequential discharge .

about selete * from table where a=XX and b=XX, Obviously it can be used (a,b) Joint index ,

about selete * from table where a=XX, It can also be used (a,b) Joint index . Because in both cases , The data in the leaf node is ordered .

however , about b Column query ,selete * from table where b=XX. You cannot use this tree B+ Tree index . Leaf nodes can be found b The value is 1,2,1,4,1,2. Obviously not orderly , So it can't be used (a,b) Joint index .

By the way:selete * from table where b=XX and a=XX, You can also use the union index , You may have questions , This statement does not conform to the leftmost matching principle . This is due to the existence of the query optimizer ,mysql The query optimizer will determine to correct this sql In what order statements should be executed most efficiently , Finally, a real execution plan is generated . therefore , Of course, we can make full use of the query order efficiency of the index , therefore mysql The query optimizer will eventually execute queries in this order .

Optimize ： Put the column with the highest selectivity at the top of the index in the federated index .

for example ： In a company with age and gender Index , obviously age Put it in front , Because there are two choices of gender, male or female , Less selective age.