Re: Shared Locator Address Pool (SLAP) protocol proposal

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

Dave,

Dave Crocker wrote:
> 
> Brian,
> 
> BEC> Dave, please look carefully at the way the NOID proposal uses
> BEC> the Flow Label as (part of) the connection ID. I had my doubts when
> BEC> it was first proposed in the design team, but it has its advantages.
> 
>  If I have suggested anything that contradicts NOID, concerning SLAP,
>  then I don't know what it is.  Please clarify.

NOID includes the flow label in the n-tuple that identifies a flow (for
lack of a better word). This isn't different from what you suggest at
10,000 meter level, but it affects some details, including the granularity
at which multihoming occurs - in fact the granularity becomes identically
the scope of a {srce, dest, flow label} 3tuple, rather than the scope of a 
{srce, dest, port, port, protocol} 5tuple. And it has the advantage of using
fields that are outside all IPSEC headers.

As far as I can tell, NOID is fully compatible with the flow label spec
(draft-ietf-ipv6-flow-label-08.txt).


> 
>  (For the total topic of multiaddressing, I want to find mechanisms that
>  do not put special or extra information into the packet header, at
>  least so it will work with IPv4, but I don't think that the SLAP
>  proposal affects this issue, one way or the other.)

This is the IPv6 multihoming WG...

   Brian