[unisog] Posting of Bandwidth Hogs

Mark Kimble mjkits at rit.edu
Tue Oct 15 15:29:05 GMT 2002


If I were a student... and this list was published ...  I would look at it
as a new Friday evening game of "Let's see who can be number-one the
fastest!"

Mark J Kimble
mjkits at rit.edu
Information & Technology Services
Rochester Institute of Technology
PGP: 80FC 8C3E 3F5B 4797 E4B4  3221 E994 2D22 1AB2 DE04

-----Original Message-----
From: Greg Schaffer [mailto:schaffer at mtsu.edu]
Sent: Saturday, October 12, 2002 10:16 AM
To: Daniel G. Epstein; Mary M. Chaddock
Cc: unisog
Subject: RE: [unisog] Posting of Bandwidth Hogs


I half-jokingly tell students to start a grass roots campaign when they
complain to me about bandwidth problems in an attempt to get their peers to
stop sharing.  Really, our BW problem has been OUTBOUND moreso than INBOUND
but that has affected all traffic (if requests are slow getting out...).  So
we have tried most of the usual tricks of the trade, but the best really
seems
to have been limiting inbound and outbound BW per connection regardless of
application. We partition the dorm traffic so ResNet can't affect the rest
of
the university (and, fortunately, most faculty/staff networks haven't (yet?)
had a P2P problem)but by limiting individual connections performance is back
to an acceptable rate.

But I think posting a top ten BW Hog list will present more problems than
solutions...

Greg

>===== Original Message From "Daniel G. Epstein" <depstein at uchicago.edu>
=====
>On Fri, Oct 11, 2002 at 05:09:59PM -0500, Mary M. Chaddock wrote:
>> We are considering the use of peer presure to help regulate the bandwidth
>> usage.
>>
>> Are any universitiies posting "Top Bandwidth Hog" lists? If so are, how
>> are you doing it? What information to you provide? Who has access to
>> it?
>>
>> Have you noticed results? What do the students think of this?
>
>Greetings,
>
>Would that count as a 'student record' under FERPA?  I guess I'd be
>afraid that someone would use it as an excuse to overreact and either
>attack the machines of the bandwidth hogs, or otherwise cause a problem
>for them.  This would result in bigger security headaches for you and
>possibly cause the bandwidth hogs to retaliate with legal action.
>IMHO, it'd be better to checkout the packet-shaping thread and see if
>you can implement something like that.
>
>Cheers,
>
>Dan
>
>--
>
>A boast of "I have been's,"  | Daniel G. Epstein
>quoted from foolscap tomes,  | Network Security Officer,
>is a shadow brushed away     | Network Security & Enterprise
>by an acorn from an oak tree |  Network Systems Administration
>or a salmon in a pool.       | NSIT, The University of Chicago
>                             | depstein at uchicago.edu
>
>GnuPG public keys available from http://pgp.mit.edu/
>or http://security.uchicago.edu/centerinfo/pgpkeys.shtml

Greg Schaffer
Director of Network Services
Information Technology Division
Middle Tennessee State University



More information about the unisog mailing list