|
|
|
Comments from IETF worker,(www.ietf.org) |
|
Jon Crowcroft (jon@cs.ucl.ac.uk) |
|
Internet Staff Member, and currently |
|
Member of Internet Architecture Board
(www.iab.org) |
|
|
|
|
Professor of Computer Science, University
College London (http://www.cs.ucl.ac.uk/staff/jon/) |
|
Worked on Internet Protocols for 20 years |
|
Funded by DARPA, HMG, and EU and industry. |
|
|
|
|
RFC1984 states IAB and IESG views on
requirements for best strength privacy |
|
Recent IETF discussion on legal
intercept/wiretap, I.e. consequenecs of CALEA |
|
Also, a few words about the non-implementabilty
of RIP |
|
|
|
|
|
IETF is representative of individual engineers |
|
IESG provide technical leadership |
|
IAB provide strategic guidance |
|
Volunteer organisations – M.O. is best expressed
as “Rough Consensus and Working Code” |
|
Very unlike ITU/ETSI (with whom we have liason
agreements) which are government or treaty orgs – we are answerable to our
membership, which is anyone and everyone. |
|
|
|
|
|
|
|
|
Next generation mobile, TV and even fixed telephone nets will all be based on IP like principles (if not n IPv4 or IPv6) |
|
To get commodity value and prices, you need to
use this technology – when you do this, it gets provided by all to
all – there are no more telco/PTT/PNO oligarchies…. |
|
|
|
|
Open Access to Strong Crypto is non negotiable
for technical reasons. It is a mandatory requirement. |
|
Specialised Technology in the form of standards
are a non requirement in IP networks. At the application level, intercept
is easy; at the IP level, impossible. |
|
|
|
|
Strong crypto is needed for operation and
maintenance – preferably public key crypto with certification chains (like
PGP) |
|
Needed for safety of critical infrastructure,
but service providers and users are indistinguishable in the Internet –
hence must be available to all. |
|
|
|
|
Dynamic Routing |
|
Multiple administrations (>40,000 ISPs) |
|
Asymmetric inter-domain (BGP) routes – most
international paths involve 3 or more administrations if you want to trace
a conversation. |
|
Triangular mobile IP routes |
|
Multicast anonymity, Etc etc etc |
|
|
|
|
First you need to capture and filter
packets on fiber in most core and
inter-tier-1 provider points at >10Gbps now. Computers don’t have fast
enough memory or buses to store more than a couple of seconds of this. In
2001, UK academic nets, for example, would require 2 terabits per second
worth of storage. You can’t filter, ship or store this: |
|
Say you need to ship it somewhere – over what
network? The net would have to be the square of its current speed – the
agencies can’t afford this … |
|
|
|
|
I can rehearse the arguments from RFC1984, and I
could re-summarise the raven debate on CALEA and IP but these are all
available online |
|
What is necessary in the EU is to inject some
technical clue into the service provider community, imho, ASAP |
|
Notes
