Re: An idea: GxSE

[Brian E Carpenter <brian@hursley.ibm.com>]

"Jon (Taz) Mischo" wrote:
> 
> On Wed, 27 Jun 2001, Brian E Carpenter wrote:
> 
> > > GxSE and NAT are mutually exclusive technologies.
> >
> > I once characterised 8+8 as "architected NAT", and both GSE and GxSE
> > fall into that category - essentially by clearly separating the mutable
> > (locator) part from the immutable (identifier) part. I suspect that
> > this separation will turn out to be a required property of the multi6
> > solution, and there are a number of ways to achieve it.
> 
> Agreed, I was saying that just to prove how opinionated I am on the
> subject.  If you read on, though, I THINK I wrote that GxSE basically
> takes lessons learned from NAT and applies them and multihoming practices
> to v6.  This may, however, be what I forgot what I was writing the above.
> 
> I'm really just trying to say that GxSE is being discussed as a way to
> avoid having to use true NAT (where the entire address is translated, as
> is the definiteion of NAT) by making a similar mechanism part of the
> addressing scheme natively.

Agreed. I'm not sure the GxSE is the best way to achieve this - once we have
the requirements agreed, we need to inspect all the possible solutions
carefully (including ones we haven't even talked about yet).

   Brian