[1724] in Commercialization & Privatization of the Internet
An interesting question from can.canet.d
daemon@ATHENA.MIT.EDU (Mark Moraes)
Mon Dec 16 03:50:05 1991
To: com-priv@psi.com
From: moraes@cs.toronto.edu (Mark Moraes)
Date: Mon, 16 Dec 1991 03:33:21 -0500
Newsgroups: can.canet.d
>From: daniel@nstn.ns.ca (Daniel MacKay)
Subject: Policy routing end-user support
Original-Message-Id: <1991Dec13.165442.14453@nstn.ns.ca>
Message-ID: <1991Dec13.165620.6932@gpu.utcs.utoronto.ca>
Apparently-To: canet-d@canet.ca
Sender: postmaster@utcs.utoronto.ca
Original-To: <canet-d@canet.ca>
Organization: NSTN Network Operations Centre, Nova Scotia, Canada
Distribution: can
Date: Fri, 13 Dec 1991 16:54:42 GMT
Approved: postmaster@utcs.utoronto.ca
Hello!
More and more often, policy routing is causing support incidents. Users,
even Internet-literate ones, will know an address, say, "dialog.com" and
will assume they should be able to connect to it. They end up very
frustrated, and call their support people for help.
The problem is, it's impossible to tell why traffic's not getting through.
Either there's a routing problem, the target machine or its router is
temporarily off the air, or there's a policy of not routing to that
network. From a philosophical standpoint, every packet is precious, and
really they shouldn't just be dropped in the trash- there should be some
kind of ICMP issued for them- the only one that seems appropriate in RFC
1009 is a "Parameter Problem" ICMP. That at would be better than nothing.
We should figure out a way to efficiently answer the question, "Is there a
policy in effect against this network?" Is there something do-able in the
near future? Could we perhaps compile a list of networks being filtered by
NSF, as a place to check when we get a "Can't Get There" support incident?
-dan
--
Daniel MacKay daniel@nstn.ns.ca
NOC Manager, NSTN Operations Centre 902-494-NSTN
Dalhousie University, Halifax, Nova Scotia, Canada
