On Fri, 2010-01-15 at 10:29 +1100, Robin Whittle wrote:
> Whether or not Name Based Sockets is a core-edge elimination
> scheme or not, I am not sure.
I'd confidently say that it is _not_ a core-edge elimination scheme.
Name-based sockets contribution is that it erases the hosts dependency
(or preference) on a given IP-address (locator). On it's own side and on
the remote hosts side.
By providing the end-hosts with the name abstraction (and locator
agnosticism), a core-edge elimination scheme might be easier to
implement or even unnecessary if the need for PI addresses could be
obsoleted completely (which I admit might be an utopian scenario ;)  )

> I will write a critique of it and
> see whether my concerns about extra host work and delays apply to
> it.
I don't think there will be any delays, all extra information is
appended to the packets "in-band". There is no 'first packet delay' due
to any pre-negotiations, the name exchange happens simultaneously, and
the individual connections are not dependent on this information being
fully negotiated before data can be exchanged. 

// Javier Ubillos

Attachment: signature.asc
Description: This is a digitally signed message part

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

Reply via email to