Smaller tables = smaller indexes. Smaller indexes also mean faster
look-ups and faster record inserts. You could eventually drop indexes on
the older tables, saving disk space (by comparison, you can't index only
part of a table). Once a table becomes so old that no updates will be
performed on it, you can even compress it saving additional disk space.
Tables that are rarely used can be moved into near-line storage (a Network
share or a SAN device) so that you save the faster local disk for the
other 95% of your queries.
Unimin Corporation - Spruce Pine
"Ronnie Sengupta" <ronnie_sengupta@stripped> wrote on 10/12/2004
> "Does splitting a large table (20 Million rows) growing at 5 million or
> a month into smaller tables improve performance given that the table can
> split in a logical way such that 95% queries don't need to look at data
> spanning across the split tables"