On Tue, 1 Mar 2005, Tom Lane wrote:
> Tatsuo Ishii <t-ishii@sra.co.jp> writes:
> > So it seems Ioannis' number was not taken immediately after a CREATE
> > INDEX operation?
>
> I would guess not, but it's up to him to say. If it is a number derived
> after some period of normal operation, then his result agrees with the
> theory that says 70% is the steady-state figure ...
yes, my number was taken after a large amount of inserts.
Your comments about the block usage in case of b-tree indexes are
absolutely interesting.
Where can i find a documentation with technical analysis for all (if
possible) of components of postgres?
All documentations that i have found are very general and refer to simple
users.
>
> regards, tom lane
>