Hi Tom, 

Sorry for the late reply. See inline:

On 12/20/10 9:39 PM, "Tom Taylor" <[email protected]> wrote:

>[PTT] One possibility is that the initial prefix potentially varies per
>customer (e.g., depending on service class) and is available from AAA.
>However, the set of possible initial prefixes has to be limited so the
>BR can recognize that it can extract a Gateway Identifier (i.e., tunnel
>identifier, see next response) from a given incoming packet.

If I read it right, the customer's prefix needs not to be embedded in the
6rd prefix. Then, how you foresee the AAA and BR exchange the customer's
prefix? In particular, this will require the AAA server to let the BR to
inject an IPv6 route per customer. This will also require the AAA server
to remember which BR the 6rd-GW must use.

>[PTT] If the tunnel is IPv6 in IPv4, the GW identifier is the compressed
>IPv4 address of the GW, as in RFC 5969. For other types of tunnel, I
>expect (but haven't worked it out in detail) that the identifier would
>be some sort of tunnel identifier. The catch would be to have an
>identifier that is common to both the GW and BR ends and fits in the
>bits available.

Got it. Depending the size of the deployment and the 6rd prefix
compression ratio, an ISP may require to deploy few 6rd domains.


Cheers,
Yiu


_______________________________________________
Softwires mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/softwires

Reply via email to