[ NNSquad ] Re: Ars technica on BGP concerns

[Ed Jankiewicz <edward.jankiewicz@sri.com>]

to quote Vint Cerf, I hope correctly, "The Internet is an experiment 
that never ended".  There has been (and still is) a lot of discussion in 
the IETF and the associated IRTF Routing Research Group, but that has 
not yet resulted in a clear resolution of the security issues resulting 
from the 20+ year old "trusting cooperative" routing model and the 
identifier/locator contradictions inherent in IP addresses.

Solutions will be evolutionary, not revolutionary, as it is hard to swap 
out the engine in flight.

On 11/23/2010 9:36 PM, Richard Bennett wrote:
> Bob's more or less right, except for the SS7 dig. BGP was designed in 
> a big hurry, to keep the Internet going post-NSFNET shutdown. It's one 
> of those things that was supposed to be improved later, and never was. 
> Its biggest problem is that it has to route IP addresses, which aren't 
> what they should be; like the URL, the IP address isn't an identifier, 
> its a locator. It was a bad move to put a locator that high in the 
> protocol stack.
>
> On 11/23/2010 6:14 PM, Bob Frankston wrote:
> > Of course. The Internet routing system is modeled after SS7 with hard 
> > outer
> > shell and a soft trusted inner core. Oh, the outer shell isn't hard. 
> > So what
> > do you expect? Like the DNS it is designed to fail!
> >
> > The answer is not more "security" in the style of ITU's hierarchical
> > authority system.
> >
> > It doesn't help that we have a hyper-complex routing system that doesn't
> > take advantage of the fact that most physical devices in the 
> > infrastructure
> > don't move around. Instead we not only assume everything is in 
> > motion, we
> > also have to route around arbitrary telecom pricing and settlement 
> > system.
> > On top of that, as per the DNS, we expect the network to keep track 
> > of the
> > end points.
> >
> > The post office knows routing a lot better than the IETF. It has a
> > (relatively) stable infrastructure and leaves it to the users to map 
> > their
> > location to a network transit point (AKA, a physical or logical 
> > address).
> > The irony is that mobility would be easier if we didn't try to collapse
> > routing and naming into one system with a single identifier that was 
> > good
> > for neither.
> >
> > This is why we need new protocols that aren't dependent on naïve 
> > trust and
> > need an infrastructure that isn't primarily about creating billable 
> > events.
> >
> >
> > -----Original Message-----
> > From: nnsquad-bounces+nnsquad=bobf.frankston.com@nnsquad.org
> > [mailto:nnsquad-bounces+nnsquad=bobf.frankston.com@nnsquad.org] On 
> > Behalf Of
> > Lauren Weinstein
> > Sent: Tuesday, November 23, 2010 20:58
> > To: nnsquad@nnsquad.org
> > Subject: [ NNSquad ] Ars technica on BGP concerns
> >
> >
> > Ars technica on BGP concerns
> >
> > http://bit.ly/fDLhHC  (ars technica)
> >
> > --Lauren--
> > NNSquad Moderator

--
Ed Jankiewicz - SRI International
Fort Monmouth Branch Office - IPv6 Research
Supporting DISA Standards Engineering Branch
732-389-1003 or  ed.jankiewicz@sri.com