Would loading the 30 million row csv file via the command line be wrapped in inside a single transaction, thus building a very rollback log? I like to break my bulk loads into nice chunks.
Also it could be the MySQL parser does a better job of optimizing the unusual select. On Sun, Feb 22, 2009 at 11:27 AM, Nicolas Williams <nicolas.willi...@sun.com > wrote: > On Sun, Feb 22, 2009 at 01:29:09PM +0100, Kees Nuyt wrote: > > > PRAGMA page_size = 20000000; /*this doesn't make any difference*/ > > > > PRAGMA page_size will only make a difference if you use it > > when creating the database (before the first table is > > created), or just before a VACUUM statement. > > Don't make it too big. 4096 or 8192 are a good start to > > experiment with. > > The hard max is 32KB, IIRC, and even that requires changing the code. > Otherwise 16KB is the max, and it works fine. > > I've wanted to make SQLite3 default to using the smaller of 16KB or the > filesystem's preferred block size. This would make SQLite3 more > efficient on filesystems like ZFS. But unfortunately the tests assume > the 1KB page size throughout, thus a number of tests fail if that change > is made. Thus for Solaris I've held back on that change for now. > > Nico > -- > _______________________________________________ > sqlite-users mailing list > sqlite-users@sqlite.org > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > -- Jim Dodgen j...@dodgen.us _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users