generators, etc....

Fri Oct 18 18:35:29 UTC 1996

In message <2131.wsimpson at greendragon.com>, "William Allen Simpson" writes:
> > From: Curtis Villamizar <curtis at ans.net>
> > If we hold BBN to the same standards that the telco industry uses to
> > come up with 99.999%, BBN was up but "a few customers" experienced a
> > localized outage.  This would be brushed off the same as the Illinious
> > AT&T fire at the POP that took out much of the Chicago suburbs for
> > about a week.  It doesn't count against the 99.999%.  (Otherwise AT&T
> > owes Chicago a couple thousand years of flawless service:).
> >
> I remember my complaint on the IETF list some years back, when
> NSFnet/ANS would routinely tout high availability, by eliminating any
> link outages from their report, as they were not under its control ...
> or route flaps greater than 5 minutes, for unknown reasons.  ;-)
> 
> WSimpson at UMich.edu
>     Key fingerprint =  17 40 5E 67 15 6F 31 26  DD 0D B9 9B 6A 15 2C 32
> BSimpson at MorningStar.com
>     Key fingerprint =  2E 07 23 03 C5 62 70 D3  59 B1 4F 5E 1D C2 C1 A2

Bill,

The routing stability reports were not a measure of availability.
They were published in the IMR (Internet Monthly Report, for those
that don't remember) and it was clearly labelled.  The section title
was "Routing Stability Measured on the T3 Network".  The very first
paragraph was:

      Internal routing stability measurements are made by monitoring
  short term disconnect times (disconnects of five minutes duration or
  less).  This is intended as a measure of overall system stability
  rather than complete connectivity.

This was started as a means to guage how well the routing software
worked, since IGP links should not go up and down a lot and IBGP
connections should never be lost unless the router dies or becomes
isolated.  It served that purpose very well.

The only incident I know of where data was removed, a commercial T1
customer with only one connection had a very intermittent circuit and
wanted it kept up while Ameritec tried to figure it out.  The problem
was in Ameritec equipment and took 2 weeks and escallation to VP level
before Ameritec solved the problem.  The fact that data was removed
was noted in the IMR report.

Curtis