RE: [nsp] Limiting bandwidth of P2P traffic

From: Andrew Fort (afort@choqolat.org)
Date: Mon May 06 2002 - 01:47:45 EDT


dek@hades.uz writes,

> rate-limiting, especially on lower speed interfaces (under
> 1Mbps) is surely to cause a lot of grief to the sessions
> you're trying to limit. CAR simply discards packets that
> exceed set limits (unless you're just marking them with lower
> precedence), and when you're doing this to a relatively small
> number of streams/sessions, will cause a lot of
> retransmissions, ultimately leading to very inefficient use
> of your bandwidth (*and* it will make your traffic profile
> very jerky).

Therefore, low-speed rate-limiting (or incorrectly small bucket sizes on
the rate-limit command, which has the same effect of causing bursty
traffic flows) is possibly perfect for the original poster's (implied)
requirement -- slowing down PTP-network traffic to the point where the
users give up and go back to doing productive work :-)

-amf



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:13:43 EDT