Re: ARPA funds Active Network research.

I Find Karma (
Sat, 26 Apr 97 14:40:49 PDT

Rohit wrote:
> [Did anyone expect otherwise now that Tenennhouse is in charge? Cool!]
> Research funded to build "active" IP networks

Man, this is DA BOMB. Active networks are THE future, although if you
talk to Rajit, he'll tell you that Caltech's Mosaic had active messaging
in 1988 and couldn't get anyone in the OS or networking communities to
listen to them because they didn't know the jargon. Hence, 10 years
later, the wheel is rediscovered.

> > An issue, however, that could delay commercial availability of the
> > technology is security. In an active network architecture, users can
> > inject customized programs into the nodes of the network. IP packets
> > are replaced with programs that are executed at each network
> > router/switch traversed.

This problem is not as bad as it first may seem.

> > "If a self-directed packet had the savvy to know where it is going, it
> > could put a lot of people out of business," said Virginia Brooks, an
> > analyst with the Aberdeen Group, in Boston.

Yes!! YES!! Kill the networking dinosaurs!!!!!! End our own private
Jurassic networking folly...

> > "I like the idea of taking network intelligence from the hardware and
> > putting it in the packet. IP is rather passive," Brooks said. "But
> > it will take a lot to figure out how you bring order to something
> > that is so democratic."

Trust building... ownership... naming... polling... markup...
automation... transactions... where have I heard these things before?
Can you say MUNCHKINS? I knew you could...

<RANT AUTHOR="Babbling Fool" ROHIT="Shut up, grasshopper!">

See, if you make IP packets active instead of passive, then you realize
that *TP... *is* the model. There is but one transport protocol needed,
and it is needed only to the effect that you get all kinds of
functionality at the task level instead of the packet-transfer level...
Hence the feature that EVERYONE can now be a router, and hence the final
nail in the coffin of end-to-end models of networking... heh heh...


Anyway, Ron wrote...

> Remember my "pregnant with itself" comments about how the network
> really isn't independent of the higher layers in the "tabletops" post?

Si. Da. Oui. YES! Ron gets it. Just like Mark got it a few weeks
back with the XML analysis. For bonus points, can you see how the two
fit together like a preordained jigsaw puzzle?

> Fwiw, stuff like this "active network" is where I believe Apollo was really
> driving at in the long term. I had some discussions with Ross MacGillivray
> at some points that were hinting at stuff like this. The "lot of people out
> of business" comment down below smells an awful lot like our traditional
> telco type operators :-).

We should take lessons where we can get em, and for this, telco have a
pretty good model.

> I give this my "post of the month" award at FoRK, where the quality varies
> from stuff this good to drivel you wouldn't believe...

\me gives you a scooby snack for wading it out.

See, we can't turn FoRK into a laserbeam of 100% clue, or we'd all go
blind. Heck, my vision is already 20/400...


Millions of peaches, peaches for free.
-- Presidents of the United States of America