Return-Path: <TMcNerney@chipwrights.com>
Received: from fort-point-station.mit.edu by po11.mit.edu (8.9.2/4.7) id MAA04513; Fri, 28 Sep 2001 12:02:49 -0400 (EDT)
Received: from intranet.chipwrights.com (ipn2-a1131.net-resource.net [216.204.61.131])
	by fort-point-station.mit.edu (8.9.2/8.9.2) with ESMTP id MAA29090;
	Fri, 28 Sep 2001 12:02:45 -0400 (EDT)
Received: by intranet with Internet Mail Service (5.5.2650.21)
	id <SVQAXPX0>; Fri, 28 Sep 2001 12:03:35 -0400
Message-ID: <93A4BBB9515ED411AE9D00508BD955A5198C31@intranet>
From: "McNerney, Tim" <TMcNerney@chipwrights.com>
To: "'John Hawkinson'" <jhawk@MIT.EDU>, Theodore Tso <tytso@MIT.EDU>
Cc: "McNerney, Tim" <TMcNerney@chipwrights.com>,
        "'sipb@mit.edu'"
	 <sipb@MIT.EDU>
Subject: RE: Who do I talk to about e19-rtr-backbone.mit.edu lossage?
Date: Fri, 28 Sep 2001 12:03:26 -0400
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2650.21)
Content-Type: text/plain;
	charset="iso-8859-1"

John,

Thanks for your input.  The more I think about a minimally acceptable
solution (and I would even be bringing this up if I didn't think it was
a real problem, and a serious degradation from my experience even 3 years
ago), is at least a solution that allows me to keep from losing my 
ssh/telnet session several times a day.  

A thought that I shared in aside conversation was that, given the bad 
"state of the net" w.r.t. "interactive" connections that process user-
typed characters, is that, perhaps TCP (and its lame retransmit policy) 
is no longer "the right thing."  People have retorted that no one cares 
about telnet/ssh anymore, but I think it is a sad state of affairs if 
"the net" has become a "ground freight only" medium at the expense of 
"air-mail letters." (i.e. lightweight, high-priority packets).  Certainly
there are interactive, multi-player games that don't work very well
when packets get dropped right-and-left, and don't get retransmitted
for several seconds.  What do these systems do to solve the problem?

--Tim

-----Original Message-----
From: John Hawkinson [mailto:jhawk@MIT.EDU]
Sent: Wednesday, September 26, 2001 7:02 PM
To: Theodore Tso
Cc: McNerney, Tim; 'sipb@mit.edu'
Subject: Re: Who do I talk to about e19-rtr-backbone.mit.edu lossage?


Theodore Tso <tytso@MIT.EDU> wrote on Sat, 22 Sep 2001
at 12:47:56 -0400 in <20010922124756.E853@thunk.org>:

> 	I think you're misdiagnosing the problem.  As others have
> said, the main problem was caused by worms running on Microsoft boxes

Remember that Tim is referring to a problem that he's seen
building for months if not years.

> No, actually it doesn't make any sense.  Remember, a router's job is
> to as quickly as possible, take the packet and move it some where
> else.  So if it's going to get handled at all, it's going to be moved
> as expeditiously as possible to the output interface.  So packets are
> *always* going to be handled as with as low a latency as possible.  A
> router can't afford to hang onto a packet, simply because it would
> take too much memory and processing power.

That's not true, Ted. There is support in high-speed routers for
doing priority queueing and related things. It incurs some
performance penalty, but typically on the order of 5-15%, which is
generally considered an acceptable price to pay.

> If you're trying to squeeze packets down a congested, slow link, such
> as down a slow 28.8k PPP connection, then it might make sense to
> reorder packets based on the port or type-of-service bits in the IP
> header, so that telnet packets get sent ahead of bulk ftp packets.

I don't think there is any in-principal problem with doing this sort
of prioritization at high speeds, no. What are you referring to?

> Finally, this sort of thing which you're describing requires making
> changes to the router's hardware and software, and as such is
> something for which SIPB (or even MIT Network Opertions) could do much
> about.  If you want to make such changes, you'd have to go work for
> Cisco and then try to exaplain to them why your idea is one which
> their thousands of network engineers haven't tried already.  :-)

Cisco's routers support this already.

--jhawk
