RE: A proposal for the improvement of the delay pools

> On Monday 15 April 2002 09:33, Chemolli Francesco (USI) wrote:
> > Also, having hierarchical pools makes hierarchy defined at pool
> > definition time rather than at ACL match time.
>
> This is true in both cases. The "hierarchy" of pools is defined at
> configuration time, bee this the pool definitions or the acl
> definitions.

Sure. It's more of a "logical" way to see things.

> The case where matrix definitions win is if you need to combine pools
> in a manner that cannot be represented by a tree. And you are
> correct, there is a few cases where this complexity is needed. But
> some additional thought need to be placed at how to configure the
> matrix.

The "matrix" view seems more natural to me, but I'd like to hear
more opinions of course. It is also arguably more expensive
in terms of performance cost, since all definition ACLs have to
be evaluated at connection setup time.

-- 
	/kinkie