Optimalizácia fullextového vyhľadávania

[PetreS]

Ahojte, vedeli by ste mi niekto poradit, ako optimalizovat PostgreSQL query, pre full textove vyhladavanie nad tabulkou, s vacsim mnozstvom dat (>=700 000)? Momentalne to riesim tak, ze nad tabulkou vytvorim GIN index pre funkciu to_tsvector, nad pozadovanymi stlpcami. Potom select vyzera nejak takto:

Kód: [Vybrat]

SELECT *
FROM table
WHERE to_tsvector(...) @@ plainto_tsquery (...)
ORDER BY ts_rank(to_tsvector(...), plainto_tsquery (...), 1) DESC

[Reklama]

[boo]

Mas tam 2x to_tsvector, aj vo WHERE a potom v ORDER co je spomalovak. Odporucam zabalit ako vnoreny select podla tohto navodu:

{code}
SELECT id, meta->>'title' as title, meta FROM (
  SELECT id, meta, tsv
  FROM data_rows, plainto_tsquery('YOUR QUERY') AS q
  WHERE (tsv @@ q)
) AS t1 ORDER BY ts_rank_cd(t1.tsv, plainto_tsquery('YOUR QUERY')) DESC;
{code}

https://blog.lateral.io/2015/05/full-text-search-in-milliseconds-with-postgresql/
https://news.ycombinator.com/item?id=9512360