Last week I was interviewing a possible new colleague for our team. During the conversation we were talking about database optimization techniques and of course indexing was one of the items on the list. While discussing this topic, the candidate made the following statement:
An index doesn’t help when using a LIKE statement in your query.
This was not in line with my idea. But he was so convinced that I decided to doublecheck. Hence this blog post…
Does a ‘LIKE’ query benefits from having an index?
Short answer: YES!
The somewhat longer answer: Yes, a LIKE statement can benefit from an index in SQL Server, but its effectiveness depends on how the LIKE pattern is constructed.
Let’s explain this with a small example.
We created a Products table and an index on the ProductName column.
Let’s now try multiple LIKE statement variations:
Suffix Wildcard (Efficient Index Usage)
This query will benefit from the index because the wildcard is at the end:
Prefix Wildcard (Inefficient Index Usage)
This query will not benefit effectively from the index because the wildcard is at the beginning:
SQL Server has to scan all rows to find any products that end with "Pro".
Infix Wildcard (Partial Index Usage)
This query may partially benefit from the index depending on the database engine's optimization:
SQL Server can use the index to find rows starting with "Lap" but needs to scan those rows to find the ones that also match "op" at the end.
No Wildcard (Full Index Usage)
This query will fully benefit from the index as it effectively performs an equality check:
SQL Server uses the index just like it would for a direct equality (=) condition.
More information
https://use-the-index-luke.com/sql/where-clause/searching-for-ranges/like-performance-tuning