Re: administrivia (on avoiding injury)

[Margaret Wasserman <mrw@windriver.com>]

Hi Joe,


>The existing draft says this:
>
>
>    Additionally, during failure events described above, multihoming
>    solutions must provide re-routing transparency for applications;
>    i.e. exchange of data between devices on the multi-homed network and
>    devices elsewhere on the Internet may proceed with no greater
>    interruption than transient packet loss during the re-routing event.
>[...]
>
>I would welcome comment on the paragraph in the existing draft.

The difference is that Brian's wording explicitly states that a
host running IPv6 based on RFC 2460 would be able to establish
new connections when one of the sites connections to the Internet
is down.

With the paragraph above, a multihoming solution could require changes 
to the existing IPv6 host stack (i.e. something newer than RFC 2460)
to achieve any connectivity when a connection is down.  It does not 
require any type of compatibility with existing IPv6 implementations.

If you combine both paragraphs, you would get:

         (1)  The solution must provide re-routing transparency for 
                 applications (possibly requiring changes to the applications
                 and/or the host stack to get this), AND
         (2)  RFC 2460 IPv6 implementations must still be able to establish
                 *new* connections regardless of which links are up 
                 or down.

I think that these are both separate, independent requirements, and I agree
with both of them.

BTW, I think that the requirement for "re-routing transparency" is one
of the places where the existing requirements docs assumes a routing
system approach (rather than an end-system approach).  It seemed clear
from discussion that the document should be editted to remove things 
that assume that the solution will be a routing solution.

Margaret