On Sun, Mar 21, 2010 at 1:30 PM, Vick Khera <vivek@khera.org> wrote:
> Like the two Scott M's recommended, figure out your usage patterns and
> partition across those vectors to optimize those searches. I would
> not worry about optimizing the insert pattern.
Note that once the partitions get small enough, on bigger hardware,
there's often little need to index at that level anymore. If you're
pulling all the records from a 50 meg db file, it'll be read in well
under a second. Even if you hit a few partitions, it's still pretty
fast since it's at worst a sequential scan, or more likely a read from
OS level cache.