3 ms·

So that's not the greatest quote but the article shows postgres having to fall back on a sequential scan for a naive COUNT(*). The author goes on to mention tha

So that's not the greatest quote but the article shows postgres having to fall back on a sequential scan for a naive COUNT(*). The author goes on to mention that as of 9.2 Postgres can do an index scan (in the context of SELECT DISTINCT): But now we come to a quirk, SELECT COUNT(DISTINCT n) FROM items will not use the index even though SELECT DISTINCT n does. As many blog posts mention (“one weird trick to make postgres 50x faster!”) you can guide the planner by rewriting count distinct as the count of a subquery To me it seems like you can trick the query planner into doing an index scan sometimes, but that it's a bit brittle. Has this improved much since?

nolevels 4y ago
I think the wiki explains it well https://wiki.postgresql.org/wiki/Index-only_scans#Is_.22count.28.2A.29.22_much_faster_now.3F https://wiki.postgresql.org/wiki/Index-only_scans#Is_.22coun... It says that index only scans can be used with predicates and less often without predicates, though in my experience I've seen the index often used even without predicates.
- inferiorhuman 4y ago
  Relevant quote: Index-only scans are opportunistic, in that they take advantage of a pre-existing state of affairs where it happens to be possible to elide heap access. However, the server doesn't make any particular effort to facilitate index-only scans, and it is difficult to recommend a course of action to make index-only scans occur more frequently, except to define covering indexes in response to a measured need And: Index-only scans are only used when the planner surmises that that will reduce the total amount of I/O required, according to its imperfect cost-based modelling. This all heavily depends on visibility of tuples, if an index would be used anyway (i.e. how selective a predicate is, etc), and if there is actually an index available that could be used by an index-only scan in principle. So yeah, I don't know if your use case is exceptionally lucky or if the documentation is just pessimistic (or both). Good to know you can coerce the query planner into doing an index scan though.

nolevels 4y ago

I think the wiki explains it well https://wiki.postgresql.org/wiki/Index-only_scans#Is_.22count.28.2A.29.22_much_faster_now.3F https://wiki.postgresql.org/wiki/Index-only_scans#Is_.22coun... It says that index only scans can be used with predicates and less often without predicates, though in my experience I've seen the index often used even without predicates.

inferiorhuman 4y ago
Relevant quote: Index-only scans are opportunistic, in that they take advantage of a pre-existing state of affairs where it happens to be possible to elide heap access. However, the server doesn't make any particular effort to facilitate index-only scans, and it is difficult to recommend a course of action to make index-only scans occur more frequently, except to define covering indexes in response to a measured need And: Index-only scans are only used when the planner surmises that that will reduce the total amount of I/O required, according to its imperfect cost-based modelling. This all heavily depends on visibility of tuples, if an index would be used anyway (i.e. how selective a predicate is, etc), and if there is actually an index available that could be used by an index-only scan in principle. So yeah, I don't know if your use case is exceptionally lucky or if the documentation is just pessimistic (or both). Good to know you can coerce the query planner into doing an index scan though.