Re: [Idr] draft-jakma-mrai-dep-00 (fwd)

On Wed, 26 Nov 2008, Pedro Marques wrote:

> Please do keep in mind that i didn't suggest that the values (0, 0) 
> are the best for your network and for the scenario that you are 
> trying to optimize. I'm merely suggesting that as general values 
> for the generic case (0, 0) are hard to beat... and they work just 
> fine.

I'm probably misunderstanding your point here, but...

While a 0 MRAI may seem very attractive for certain examples, the 
studies with simulations of larger-scale topologies (Griffin and 
others) suggests some appreciable amount of an MRAI is required to 
avoid an explosion in the number of messages. Additionally these 
simulations show convergence time /increases/ proportionally with the 
number of messages (as the MRAI is lowered from optimum toward 0) - 
presumably due to processing and messaging overheads.

(Course, as someone pointed out to me in private, even a nominally-0 
MRAI still implies some amount of per-node processing delay).

So it seems like the studies available suggest it could be quite bad 
if all speakers on the internet were to switch to 0/0, even where 
such values might be of benefit to subsets of the internet.

On a different note:

While it seems everyone is agreed the existing 30s/5s MRAI 'SHOULD' 
values are too high and should be revised in some way, it's also 
apparent there is no agreement on what those specific values should 
be (some think 0/0, some max-5/max-1, some think those values aren't 
conservative enough).

So it seems like just removing/deprecating any specific recommended 
values is the most we can hope to achieve consensus on.

regards,
-- 
Paul Jakma	paul@clubi.ie	paul@jakma.org	Key ID: 64A2FF6A
Fortune:
"Aww, if you make me cry anymore, you'll fog up my helmet."
-- "Visionaries" cartoon
_______________________________________________
Idr mailing list
Idr@ietf.org
https://www.ietf.org/mailman/listinfo/idr