>I can imagine having a "this is short-lived" socket option, and if the app
>sets the option then the stack would prefer care-of addresses over home
>addresses instead of the other way around. This is the app telling the stack
>that your condition (b) holds. I don't see the importance of condition (a).
>
>Assuming you send a binding-update in the SYN, then communication using the
>home address will be relatively efficient - just a little more overhead in
>the packets, but no extra packets and no packets going through the home
>agent.
Mobile-ip6 requires only one thing to correspondent node:
honor home address option.
Spec does not require every correspondent node to have binding cache,
so binding update in SYN may not work and result in triangle routing.
Therefore, I still think this is valuable for mobile node to be able
to use care-of as TCP/UDP source address, for short-lived sessions.
For me, short-lived sessions include:
- local services in remote site the mobile node is visiting
access to web cache in remote site it visits, access to printer
- use ubiquitous services in remote site (DNS)
>The main exception would be when you are dealing with a server that handles
>lots of short-lived connections - it's binding cache might not work well. So
>for your short-lived connections to such a server, you'd want to use your
>care-of address instead of your home address.
I suspect that busy web servers will not have (or will not use)
the binding cache. The server will not be able to maintain it.
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]
--------------------------------------------------------------------
