IEEE: A radical new router ?

[Robert E. Seastrom]

andre kesteloot <andre.kesteloot at verizon.net> writes:

> http://www.spectrum.ieee.org/computing/networks/a-radical-new-router

Ah yes, Larry Roberts espousing flows in the core (again).  Been
there, done that, got the T-shirt.  Cisco 1995 I believe it was - a
bitter vintage with a distinct note of panic at being behind the power
curve.  Perhaps mo would care to elaborate in his inimitable fashion.

Not only has this been shown to be in practice a Bad Thing (in the
core of the Internet, you don't necessarily see discernable "flows"
and the opposing directions of a session are often taking entirely
different paths so you'd see only one side anyway), but if one
actually tried to pay attention to groups of packets and create/tear
down state for special treatment as groups come and go, at higher
speeds that ends up being *all* one is doing.  Opens up a tremendous
vector for attacks of various types; even unintended side effects of
self-propagating software that scans a lot can give you a pretty
serious pain.

I don't think this is going anywhere - I don't know anybody who's
spending money on Anagran boxes just as I never actually saw Caspian
kit in the wild.

-r