All fantastic suggestions. The EXPLAIN results showed that changing to logging_userindex had potentially the largest impact (reducing the rows to be retrieved from 10mil to 5mil). I am rerunning the query with all suggested recommendations, and will report back.
On Sat, Dec 30, 2017 at 1:32 PM, Dennis Tobar <dennis.to...@gmail.com> wrote: > Replace count(*) with count(1) in the subquery. It could help (?) to > improve the performance. > > Dennis Tobar Calderón > (Enviado desde un móvil) > > El 30 dic. 2017 3:10 p. m., "John" <phoenixoverr...@gmail.com> escribió: > >> I would also find the first log of 2017 and use that instead of the >> timestamp >> >> _______________________________________________ >> Wikimedia Cloud Services mailing list >> Cloud@lists.wikimedia.org (formerly lab...@lists.wikimedia.org) >> https://lists.wikimedia.org/mailman/listinfo/cloud >> > > _______________________________________________ > Wikimedia Cloud Services mailing list > Cloud@lists.wikimedia.org (formerly lab...@lists.wikimedia.org) > https://lists.wikimedia.org/mailman/listinfo/cloud >
_______________________________________________ Wikimedia Cloud Services mailing list Cloud@lists.wikimedia.org (formerly lab...@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud
