CODEWITHSUNDEEP
sqlpostgresqlintersectionpostgresql-performance
Antonio Gamiz Delgado
Antonio Gamiz Delgado

Reputation: 2103

Intersection of SQL statements efficiently

I'm trying to solve a SQL problem where I need to select the data in a certain intersection. This is more or less my query:

SELECT id FROM some_table WHERE
id IN (SELECT id FROM other_table WHERE _id=1 AND some_value >= 1 AND some_value <= 10) AND
id IN (SELECT id FROM another_table WHERE _id=2 AND some_other_value >= -65 AND some_other_value <= -2)

The problem with that query is that it does not return the intersection because the _id field is different both subqueries. So that query will return an empty result always.

If I use OR instead of AND to "intersect" the subqueries, then the intersection is also not returned.

I also do not want to use INTERSECT because that's kind of slow and this is just a very reduced example of a very large query, so speed is really important.

So, my question is, is there any way to do this intersection as fast as possible? Have in mind that even though in the example only appears two subqueries, in my real use case the number of subqueries can be larger.

Upvotes: 0

Views: 957

Answers (1)

xQbert
xQbert

Reputation: 35323

I'm still unclear on what you're after w/o sample data or expected results, but would either of the following options work?

using INNER JOIN

SELECT id 
  FROM some_table A
 INNER JOIN (
       SELECT id
         FROM other_table
        WHERE _id = 1
          AND some_value >= 1
          AND some_value <= 10
 ) B on A.ID = B.ID
 INNER JOIN (
       SELECT id
         FROM another_table
        WHERE _id = 2
          AND some_other_value >= -65
          AND some_other_value <= -2
 ) C ON C.ID = A.ID

using INTERSECT

SELECT id
  FROM some_table
INTERSECT
SELECT id
  FROM other_table
 WHERE _id = 1
   AND some_value >= 1
   AND some_value <= 10
INTERSECT 
SELECT id
  FROM another_table
 WHERE _id = 2
   AND some_other_value >= -65
   AND some_other_value <= -2


Then, I suppose we could use `EXISTS` and correlated subquery—instead of `IN`. This can be faster, as `EXISTS` can escape early whereas `IN` cannot.

Upvotes: 1

Related Questions