Home/ Questions/Q 6757033
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-05-26T13:35:53+00:00

We observed that, while joining a table with a multi-column PK to itself: a

  • 0

We observed that, while joining a table with a multi-column PK to itself:

  1. a join using all PK columns ran in .5s
  2. a join using all but one of the PK columns ran in 42-44s

Other, maybe extraneous information:

  • all PK columns had btree indexes
  • the column that we left out in the second case was a BOOLEAN (actually tinyint(1)) – so the result set was twice as large as in the first case
  • this was observed for both inner join and left join
  • no non-PK columns were used in the joins
  • running the query repeatedly did not result in a large decrease in execution time (maybe 2s)

Why was the difference in execution time so large?

Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    There is a unique index placed on a (set of) PK columns. View the explain plan to verify, but it is likely that the index is only used when joining on all PK columns.

    • 0