On Wed, Oct 2, 2013 at 9:01 AM, David Rysdam <drysdam(at)ll(dot)mit(dot)edu> wrote:
> I had some vague notion of tables "doing work" but really if it can load
> one
> partition into RAM and get most of my hits from there, it'd be a big
> win.
The same concept applies to the frequently-used indexes on that partition.