On Tue, Feb 10, 2009 at 6:37 PM, Brian Aker <[email protected]> wrote: > Hi! > > On Feb 10, 2009, at 5:12 PM, Padraig O'Sullivan wrote: > >> reason why hash joins are not implemented in Drizzle? I would have > > No one has done it :) > > Its on my radar to do it, but I don't have the time to look at it. I'd > welcome patches for it. > > Despite what Mark believes, I think that performance of RAM vs SSD is such > that it still makes sense (especially when rows have been rendered into > temporary tables). > > Cheers, > -Brian > >
Of course we need this feature and thousands of others. We can't afford to build all of them. The issue is whether hash join is so much faster than nested loops join + batch key access and optionally SSD that it is worth the effort to build it and then to deal with more bugs when the optimizer chooses the wrong join method. An additional factor in this is that we now have more options for doing large scale query processing outside of a DBMS (Hadoop, Pig, ...). > -- > _______________________________________________________ > Brian "Krow" Aker, brian at tangent.org > Seattle, Washington > http://krow.net/ <-- Me > http://tangent.org/ <-- Software > _______________________________________________________ > You can't grep a dead tree. > > > > > _______________________________________________ > Mailing list: https://launchpad.net/~drizzle-discuss > Post to : [email protected] > Unsubscribe : https://launchpad.net/~drizzle-discuss > More help : https://help.launchpad.net/ListHelp > -- Mark Callaghan [email protected] _______________________________________________ Mailing list: https://launchpad.net/~drizzle-discuss Post to : [email protected] Unsubscribe : https://launchpad.net/~drizzle-discuss More help : https://help.launchpad.net/ListHelp
