SQL JOIN vs IN performance?

Generally speaking, IN and JOIN are different queries that can yield different results.

SELECT  a.*
FROM    a
JOIN    b
ON      a.col = b.col

is not the same as

SELECT  a.*
FROM    a
WHERE   col IN
        (
        SELECT  col
        FROM    b
        )

, unless b.col is unique.

However, this is the synonym for the first query:

SELECT  a.*
FROM    a
JOIN    (
        SELECT  DISTINCT col
        FROM    b
        )
ON      b.col = a.col

If the joining column is UNIQUE and marked as such, both these queries yield the same plan in SQL Server.

If it's not, then IN is faster than JOIN on DISTINCT.

See this article in my blog for performance details:

• IN vs. JOIN vs. EXISTS

Funny you mention that, I did a blog post on this very subject.

See Oracle vs MySQL vs SQL Server: Aggregation vs Joins

Short answer: you have to test it and individual databases vary a lot.

That's rather hard to say - in order to really find out which one works better, you'd need to actually profile the execution times.

As a general rule of thumb, I think if you have indices on your foreign key columns, and if you're using only (or mostly) INNER JOIN conditions, then the JOIN will be slightly faster.

But as soon as you start using OUTER JOIN, or if you're lacking foreign key indexes, the IN might be quicker.

Marc

This Thread is pretty old but still mentioned often. For my personal taste it is a bit incomplete, because there is another way to ask the database with the EXISTS keyword which I found to be faster more often than not.

So if you are only interested in values from table a you can use this query:

SELECT  a.*
FROM    a
WHERE   EXISTS (
    SELECT  *
    FROM    b
    WHERE   b.col = a.col
    )

The difference might be huge if col is not indexed, because the db does not have to find all records in b which have the same value in col, it only has to find the very first one. If there is no index on b.col and a lot of records in b a table scan might be the consequence. With IN or a JOIN this would be a full table scan, with EXISTS this would be only a partial table scan (until the first matching record is found).

If there a lots of records in b which have the same col value you will also waste a lot of memory for reading all these records into a temporary space just to find that your condition is satisfied. With exists this can be usually avoided.

I have often found EXISTS faster then IN even if there is an index. It depends on the database system (the optimizer), the data and last not least on the type of index which is used.