I do not see the difficulty. We already have the IPv6 mechanisms to
advertise the prefixes. And we have the IPv6 mechanisms for hosts to
combne the prefixes with teir IDs.
We also have the dynamic DNS mechanisms (with security) needed to
advertise the results.
With ILP, these combinations can be changed during sessions, and traffic
can change paths during sessions without impact.
With the current architecture, sessions can not change paths, and
changes to the connectivity are hard to discover or utilize.
Thee are other multi-homing problems that are not solved. Working out
how to manage prefix assignments in an enterprise when external
assignments change is one example of such issues (LISP takes a different
tack, and thus the costs and benefits are different.)
Yours,
Joel
On 11/12/2012 6:59 PM, Dae Young KIM wrote:
On Tue, Nov 13, 2012 at 8:40 AM, Noel Chiappa <[email protected]
<mailto:[email protected]>> wrote:
> From: Dae Young KIM <[email protected] <mailto:[email protected]>>
> This multiplicity of the locators, although the ID is single
... also
> affects DNS entries.
Not necessarily. If the DNS only contains IDs (which is the way some
designs
work), then there no DNS/locator issues.
The description on servers in Sec. 4.1 of ILNP, p. 20, implies that.
The DNS concern is, however, a secondary thing. That a host has to be
associated with multiple PA locators inserted by multiple ISPs at ILNP
multi-homing is the real problem I'm pointing to. This is exactly the
same situation you might require with the current IP.
Where's the improvement on multi-homing? If there's no such improvement,
have we solve the table explosion problem?
--
DY
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
