FYI

Stefan Savage (savage@cs.washington.edu)
Tue, 16 Nov 1999 13:35:32 -0800

I have a neat graph from the "Network Reliability Steering Committee" which
is an telco inter-provider organization that is charged with understanding
the sources and causes of network failures in that world.

Anyway, of 107 Major Fiber Cable Dig-ups (must affect > 30,000 customers for
more than 30 mins to be reportable... so this doesn't include SONET heals)
in the last quarter, the responsible parties were:

Construction Company: 18%
Municipality: 17%
Utility: 16%
Railroad: 15%
Property owner: 14%
State Government: 8%
Other: 7%
Telco: 5%

In terms of FCC reportable failures (I believe NOT included fiber digups)
there were 46, of which:
17 were facility failures, 9 were local switch failures, 7 were tandem
switch failures (tandem switches link together other switches... class 4
tandems link together local switches, class 3 switches tie together class 4
tandem switches, etc... up to class 1 international gateways... generally
only one per country), 4 were central office power failures, 7 were common
channel signaling failures, 2 were digital cross-connect systems failures.
Of these, 25 were attributed to procedural errors (human fuckups).

Finally, the mean time between outages is on the order of 2 days and the
median duration of outages is a bit under three hours.

Finally, they have an internet report which shows that ISP dial-up traffic
is expected to dominate the peak-load engineering at both the central office
and at the trunk in the next 2-3 years.

Interesting stuff.

- Stefan