In an attempt to throw the authorities off his trail, threshar@torgo.978.org (Jeff) transmitted:
> One area though is fast recovery - Reloading a multi-GB db from a
> pg_dump is painful, especially if you have foreign keys. Lots of
> sort_mem helps.
A very interesting approach on a multi-cpu system would be to restore
multiple indices concurrently when working on a single table. The
locality of reference should allow this to be handled quite
efficiently, as data blocks pulled in for one index should shortly be
used for others.
This ought to get decent parallelism.
--
wm(X,Y):-write(X),write('@'),write(Y). wm('cbbrowne','cbbrowne.com').
http://www3.sympatico.ca/cbbrowne/linuxdistributions.html
"To conquer the enemy without resorting to war is the most desirable.
The highest form of generalship is to conquer the enemy by strategy."
-- Sun Tzu, "The Art of War"