Network Event Reporting Facility

Sean Donelan SEAN at SDG.DRA.COM
Fri Jun 12 03:06:59 UTC 1998


CNN has been reporting on the MCI fiber cut in Manhattan.  Since various
other carriers lease each other's facilities, it has also been affecting
worldcom, uunet and others.  Sprint and AT&T claim they haven't been
affected.  The last report I got around 6pm, MCI hadn't located the
actual fiber cut or cause; only the general vicinity.  I have not found
an official statement on this outage on the MCI web site.

I had posted the following a couple of weeks ago to the
outage-discuss at academ.com mailing list, but other network operators may
be interested in methods for keeping their customers informed about
operational issues (besides relying on CNN).

====================================================================

I'm off charging at windmills again.  I've written up a protocol so
network providers can inform their users (and maybe other providers)
about significant network events in a distributed, robust manner.  If
you recognize parts of this, I have stolen the best ideas from half-a-dozen
other protocols, News, NTP, SDR, and even OSPF and RIP.

Since communication satellites, pagers, SS7 networks, SONET rings and so
forth have been known to fail from time to time, simply relying on any
single method for notifying users about problems isn't sufficient.  The
net has suffered from the 'flash-crowd' effect when everyone tries to
check centralized information sources such as web servers.

I have attempted to address some of the issues in other notification
methods.  Namely I've tried to give provider control and authentication
of messages, as well as giving enough information so end-users to
filter messages of interest to them.  I've tried to assume if someone
can abuse the system, they will; so things are a bit paranoid.  Stale
messages remain a problem of any notification system, so I tried to
give providers an explicit mechanism to cancel messages within their
own network.

I would appreciate any interested persons looking over my current attempt
at writing a draft at http://dranet.dra.com/nerf.html.  I have also set
up a mailing list nerf-l at listserv.dra.com (subscribe by sending a 
message to listserv at listserv.dra.com) for interested people.  
-- 
Sean Donelan, Data Research Associates, Inc, St. Louis, MO
  Affiliation given for identification not representation




More information about the NANOG mailing list