> So are you _really_ sure you want to drop RH0 from the IPv6 protocol
> spec, given that the practicalities of a general use 'replacement'
> header is going to present some fascinating transition issues?

I suspect code changes are necessary in any case, completely
independent of whether RH0 or RHx is going to be employed
or what the IETF does.

First, given the possibility for amplification, implementations
would be well advised to at least add some checks and limits
on RH0, even if it would not be deprecated. Second, a number
of implementations already have turned the feature off, which
obviously means that we cannot expect a random node in the
Internet to always respect anyone's request to perform a RH0
operation.

But, at the end of the day, I'm not sure the above is a big
problem.

Please remember that RH works in an end-to-end fashion;
only the source and current destination of the packet needs to
process it. Not routers in between.

There have been examples where new RH types
have been specified, implemented, and used. In Mobile IPv6,
for instance, RH2 is implemented by the participants. This
is easy, as only the participants need to implement it,
and RH2 is a MUST-to-implement in the specifications
for any of the participant nodes in Mobile IPv6.

More generally, if one wants to provide a new feature based
on routing header, what we need is

- Implementation in nodes that want to use RHx for something.
  This should be easy.

- Implementation in other, consenting nodes that appear in
  a non-final destination of the routing header.

  This is not needed in all cases, e.g., not in the Mobile Ipv6
  case.

The second item is harder, but I'm not sure its an unreasonable
requirement. After all, as a manager of a network you probably
want to be in control of where re-forwarding to a new address
is used. And you probably are in control of that, already...

Jari



--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to