Sargable
In relational databases, a condition (or predicate) in a query is said to be sargable if the DBMS engine can take advantage of an index to speed up the execution of the query. The term is derived from a contraction of Search ARGument ABLE. It was first used by IBM researchers as a contraction of Search ARGument, and has come to mean simply "can be looked up by an index."1
For database query optimizers, sargable is an important property in OLTP workloads because it suggests a good query plan can be obtained by a simple heuristic2 matching query to indexes instead of a complex, time-consuming cost-based search, thus it is often desired to write sargable queries. A query failing to be sargable is known as a non-sargable query and typically has a negative effect on query time, so one of the steps in query optimization is to convert them to be sargable. The effect is similar to searching for a specific term in a book that has no index, beginning at page one each time, instead of jumping to a list of specific pages identified in an index.
The typical situation that will make a SQL query non-sargable is to include in the WHERE clause a function operating on a column value. The WHERE clause is not the only clause where sargability can matter; it can also have an effect on ORDER BY, GROUP BY and HAVING clauses. The SELECT clause, on the other hand, can contain non-sargable expressions without adversely affecting the performance.
Some database management systems, for instance PostgreSQL, support functional indices. Conceptually, an index is simply a mapping between a value and one or more locations. With a functional index, the value stored in the index is the output of the function specified when the index is created. This capability expands what is sargable beyond base column expressions.
- Sargable operators:
=, >, <, >=, <=, BETWEEN, LIKE, IS [NOT] NULL, IN - Sargable operators that rarely improve performance:
<>, NOT, NOT IN, NOT LIKE