Difference between GiST and GIN index

Question

I am implementing a table that has a column with a data type of tsvector and I am trying to understand what index would be better to use?

GIN or GiST?

In looking through the postgres documentation here I seem to get that:

• GiST is faster to update and build the index and less accurate than gin.

• GIN is slower to update and build the index but is more accurate.

OK, so why would anybody want a gist indexed field over gin? If gist could give you the wrong results? There must be some advantage (outside performance) on this.

Can anybody explain in layman's terms when I would want to use GIN vs. GiST?

Polymorphonuclear

Always provide your version of Postgres. GIN has received major improvements in Postgres 9.4 – Acrogen

Answer 1

I don't think I could explain it better than the manual already does:

In choosing which index type to use, GiST or GIN, consider these performance differences:

• GIN index lookups are about three times faster than GiST

• GIN indexes take about three times longer to build than GiST

• GIN indexes are moderately slower to update than GiST indexes, but about 10 times slower if fast-update support was disabled [...]

• GIN indexes are two-to-three times larger than GiST indexes

Link and quote refer to the manual for Postgres 9.4. Size and performance estimates seemed slightly outdated already. With Postgres 9.4 the odds have shifted substantially in favor of GIN.
The release notes of Postgres 9.4 include:

• Reduce GIN index size (Alexander Korotkov, Heikki Linnakangas) [...]

• Improve speed of multi-key GIN lookups (Alexander Korotkov, Heikki Linnakangas)

Size and performance estimates have since been removed from the manual.

Note that there are special use cases that require one or the other.

One thing you misunderstood: You never get wrong results with a GiST index. The index operates on hash values, which can lead to false positives in the index. This should only become relevant with a very big number of different words in your documents. False positives are eliminated after re-checking the actual row in any case. The manual:

A GiST index is lossy, meaning that the index may produce false matches, and it is necessary to check the actual table row to eliminate such false matches. (PostgreSQL does this automatically when needed.)

Bold emphasis mine.

Acrogen

I believe you meant "You never get wrong results with a GIN index", right? – Vagrancy

@IamIC: You never get wrong results with either GIN or GiST. But I am specifically addressing GiST in the answer because the OP had a wrong impression there. – Acrogen

Understood. That makes sense. – Vagrancy

If there are more reads than records in your table, you need to use GIN. – Anarchism

I think the point GIN indexes are two-to-three times larger than GiST indexes is not valid as I have tested with few large tables and found GiST index taking up more space than GIN and BTree indexes – Irreplaceable