HK <harikrishnan@midascomm.com> writes:
> time in ms per query retrieval
> py key non py key
> ------ -----------
> 2000 15.8 15.7
> 10000 71.2 70.7
> 25000 174 173.5
> 100000 4319 3417
> As u see the non primary key retrieval is faster. AFAIK accessing primary
> key should be faster. Can anybody throw light on why is this anamoly.
There's nothing magic about primary keys in Postgres; I'd expect the
results to come out the same.
The first three of these look like they're indeed the same, to within
measurement error. Not sure about the last one; perhaps you had some
effect of physical order in the table agreeing with the ordering of the
non-primary key?
regards, tom lane
