On Tue, Apr 20, 2010 at 8:44 AM, Toni Stoev <[email protected]> wrote:
> On Monday 19 April 2010 at 05:32:31 Robin Whittle sent:
>> Short version:  RRG participants should contribute to the debate by
>>                 having a go at writing their own preferred goals and
>>                 Recommendation and/or by critiquing other people's
>>                 attempts to do so.
>
> 1. identity/location separation, for the sake of DNS, mobility, session 
> management; done by introducing identifier.

52% of the Internet traffic is using HTTP, there you have a cookie
that can be used as a session identifier. MPTCP and SCTP have their
verification tag/token, these can be used as a session identifier.
I'm not that convinced that we need a generic identifier to solve the
routing issue - the mobility issue is more or less an application
issue, you have to take into account several factors such as low and
choppy bandwidth, the IP address might change etc. So the application
should use HTTP's cookie or leverage the transport protocols session
identifier.

With an identifier you can have the same checksum for the transport
protocol though your locator changes - but the reality is that there
is so much NAT out there that the Internet services no longer care if
the checksum of transport protocol changes.

Rendezvous services can be found for "free", e.g. Skype, pee-to-peer,
IM-services, SIP proxy/registrar etc
To use a generic identifier and to protect it from theft  you should
use DNSSEC, but that will come with an extra cost
http://www.networkworld.com/news/2010/032310-domain-name-registars-lagging.html
Will the average John Doe pay extra to protect his generic identifier
or use "free" identifiers that are bundled together with services he
uses??

Technically the locator/identifier separation is great, but I'm afraid
it isn't disruptive compared to the existing architecture we have -
think that MPTCP's token will just be good enough to provide mobility
for John Doe.

-- patte
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg

Reply via email to