Re: I-D ACTION:draft-ietf-shim6-applicability-01.txt

[Brian E Carpenter <brc@zurich.ibm.com>]

marcelo bagnulo braun wrote:
Marcelo,

...
> So, bottom line is what a HBA/CGA dhcp extension would be useful for and 
> what scenarios/use cases are there for this?

I think it gets pretty hard when you include CGA in the problem. I'm
concerned about a site of medium size that wants to manage its
nodes proactively and therefore wants to use DHCPv6, but also wants
to profit from shim6, with the level of security offered by
HBA. That seems like a real scenario to me.


> > > 5.3.  Shim6 and SCTP
> >
> > ...
> >
> > >    ...It is
> > >    recommended that Shim6 is not used for SCTP sessions, and that path
> > >    maintenance is provided solely by SCTP.
> >
> >
> > What is the mechanism, i.e. how does SCTP tell shim6 to do nothing?
>
> well, i think we discussed about this and my understanding was that 
> there was no need for an explicit notification by the sctp to the shim 
> (this is implementations specific proabbaly...)
>
> That it would be possible for the shim layer to simply not create any 
> context for sctp sockets. In order to do that, the shim6 layer need to 
> have access to the socket information (which i think is implementation 
> specific)
>
> Besides, it would also be possible to define a DONOT_SHIM socket option, 
> but this would require update to use such mechanism

I think you need to say this in the draft. You are assuming that
a stack that supports SCTP "knows" when a socket is for SCTP, and that
needs to be confirmed. If not, a socket option seems the only way,
and IMHO we should say so.

   Brian