Well, since you're grouping, you don't need the distinct on (it's
sort of assumed by group by - you're only going to get one group
for each distinct value set of your grouped columns). In my
explain, this cuts out a unique step on my small tables, not
sure how significant that's going to be on an already unique
set of columns though.
On Fri, 19 Jan 2001, Konstantinos Agouros wrote:
> Hi,
>
> I have a table with the fields data(timestamp), region and url (varchar)
> This table has 4.5mio entries a day. To reduce them I use the following
> insert-statement.
>
> insert into msreduce select distinct on (region, url) '1-16-2001',region,url,count(client) from microsoftlog where
date> '1-16-2001' and date < '1-17-2001' group by region,url
>
> My question: Could this be optimized?