Marcello.
However strange it may seem, I now put the APBP approach in stanby,
probably to finally simply give it up.
Reasons:
- The tunnel-to-NAT44 approach is IMHO so advanced, and probably "good
enough", so that the work on APBP, even if pertinent, would arrive too
late.
- Some server applications are reported to require that related TCP
connections do have the same client IP address. This conflicts with
the possible load balancing between several APBP servers, one of their
nice scalability features.
- Some of the underlying concepts of APBP may be better applied in a
different context (a combination of multi-homing and IPv6 private
addressing, but this is another story, not related to your draft).
I therefore no longer suggest that having APBP mentioned in your next
version(s) is desirable.
In any case thanks for your openness to reflect received comments on it.
Regards.
Rémi
marcelo bagnulo braun - Le 6/19/08 7:59 PM :
In the next version, i will try to reflect the received comments indeed
Rémi Després escribió:
Please remember what Teemu Savolainen wrote to you in
http://psg.com/lists/v6ops/v6ops.2008/msg00800.html
Also you can note what Brian Carpenter answered to Iljitsch van
Beijnum, in
http://www.ops.ietf.org/lists/v6ops/v6ops.2008/msg00876.html