>Before all, thanks for clarifying the anycast subject !
>
>For TCP connections, or UDP with suspicious clients, it seems to me the
>pb is quiet the same as for mobile node acces, hence why not use the
>same solution ?
>The anycast server (who knows it owns an anycats @) can play the role of
>a mobile, using its anycats @ as Home-@, and one of its current @ as
>Co@. By using the Home Address option the UDP client won't know the
>difference, and if necessary (for TCP),
the document (01 revision) has a comment about the use of home
address option for overriding source address. it was suggested
(by Steve Deering and other folks) that, home address option should
have the same restriction as the IPv6 source address field,
so use of multicast/anycast is discouraged.
the problem if we use multicast/anycast as the source is, we cannot
identify single node as the source.
we need a consensus on this, and also we need explcit notice on spec.
>it can also send Binding Update
>options, to be sure to be the one and onlmy server speaking with this
>client.
>Then the client would not have to know wether the @ its using is anycast
>or not.
it is not mandatory for an IPv6 nodes to have binding cache, so
binding update may not work.
also, how can you establish IPsec keys while using anycast address?
itojun
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
