Re: classifying P2P traffic

[Valdis Kletnieks <Valdis.Kletnieks () VT EDU>]

On Tue, 29 Jan 2008 19:04:41 CST, John Kristoff said:
> On Tue, 29 Jan 2008 09:18:55 -0600
> "Julian Y. Koh" <kohster () NORTHWESTERN EDU> wrote:
>
> > dynamic subpartitions for our dorm/wireless/VPN IP ranges to limit
> > unclassifiable traffic to 512Kbps per host based on IP address.  But
> > overall it seems to be working quite well with that arrangement.
>
> Does anyone just do that, per /32 (or something slightly larger),
> limiters or dropping knobs and not bother trying to classify the app?

Locally, we just count octets per switch interface, and if the upstream
traffic on a dorm port goes over a certain limit per 24 hours, we apply a
rate limit sufficient for most use of the net (checking e-mail, web surfing)
since we do *not* limit downstream traffic.  Protocol doesn't enter into it
at all, just the octets, and we don't bother trying to do up-front filtering
of "illegal" content - they got the copyright lecture at orientation, and
if they choose to not listen, the vast majority of them are legally adults,
so if/when the RIAA or MPAA send a complaint, it's not our problem...

Attachment: _bin
Description: