-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Matthew Toseland wrote:
> One of the receiver windows? As I recall we were going to use token
> buckets for fair sharing (on _incoming_ requests).

Right. The receiver window we advertise to each peer tells it how many
tokens are in its bucket.

> Additionally RejectedOverloads are no longer propagated back to
> source. Load propagates back to source not by RejectedOverload's being
> forwarded to the original sender (as now), but by nodes punishing
> flooders

I hope so, but I'd like to see how the simulations turn out before
committing myself either way - if we can only manage to restrict
flooders to an even share of their neighbours' bandwidth,
RejectedOverloads might still be useful for getting well-behaved nodes
to slow down.

Cheers,
Michael
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)

iD8DBQFEkG4Hyua14OQlJ3sRAteLAKDnr4JgHAGRlNOQZWOTo5AP5XpOBACfZ3CO
A2omcMUVL369flVmQHQZuvU=
=xnoG
-----END PGP SIGNATURE-----

Reply via email to