|Subject||Re: commit after insert is very slow|
simply in some area (province or region) where you especially look in the archived transaction ... i already do partitionning (in 40 tables), but partionning have also a cost when you want to mix/union the result :(
--- In email@example.com, Dimitry Sibiryakov <sd@...> wrote:
> > the table have more than 10 000 000 records (actually many more)
> > now we need to retrieve any kind of subset in any of the previous available order by, and this in lower than 100 ms !
> Curiosity: what kind of area has more than 10 000 000 houses? What
> will happen at 101st millisecond? They will burn?
> IMHO - you should think about partitioning.
> SY, SD.