Outage scale

Reacting to "2000 ds3s" as a large number.
We are approaching the day when we will see a million ds3 backhoe cut. At 45
terabits, this is within the range of lab WDM systems (>1Tb per strand.) We
are still years away from seeing these widely deployed in the ground, but the
scale of outages is going to go way up.
Oddly, it is also possible that the impact of such an outage may go down. If
the bandwidth prices get cheaper, then it becomes more reasonable to have high
levels of diversity. We all understand the difficulties in keeping that
diversity real in the current world.

Doubtful. Diversity requires the one thing which they aren't making more,
land. The general pattern has been to put more bandwidth into fewer paths,
roll everything onto the new technology, remove the old "obsolete" path
from service and use it for something completely different. Putting in a
single OC-192 doesn't cost much more than putting in a OC-3 fiber. But if
you are only using 14% of your OC-192 path, why would any bean counter let
you put in a second OC-192? Just put your protect circuit in the unused
part of your existing OC-192 path (duh!)

Bell Atlantic has a great FCC outage report where they say there are no
recommendations for SONET diversity. Hanging both halves of the ring from
the same pole doesn't violate any SONET constraints. Southwestern Bell has
almost a half dozen different outages this year isolating dozens of towns in
most cases due to single transmission facility failures in their new fiber
optic lines.

Heck, the last New Jersey outage even took out some of the state's NAWAS
(Nuclear Attack Warning) circuits in northern New Jersey. Fortunately the
Russians are our friends now.

I think diversity will go down as one of those great unsolved problems, along
with the Halting Problem.