3 ms·
2c: in most cases, a Postgres index is (performance wise) identical to a "real column" just without the name, and indeed you have more control over the data str
by asah 5y ago
2c: in most cases, a Postgres index is (performance wise) identical to a "real column" just without the name, and indeed you have more control over the data structure for an index, and indexes can include lossy compression (e.g. partial indexes) unlike a real column...
(obviously, the challenge is getting pg to use your index for a given query... for that, I like to use VIEWs to help query authors write queries that use the indexes)
- saurik 5y agoYeah. This wasn't the case a while back--PostgreSQL was missing support for what I think people call covered indexes, and so would only use the index to filter a set of candidate rows and then still had to go digging around in the real tuple to verify it matched--but I am pretty sure they changed that a while back.