Re: [v6ops] PI [ULA draft revision #2 Regarding isolated networks]

[Mark Andrews <[e-mail]>]

In message <[e-mail]>, Philip Homburg writes:
> In your letter dated Sun, 1 Jun 2014 16:10:18 -0400 you wrote:
> >On Jun 1, 2014, at 1:36 PM, Lorenzo Colitti <[e-mail]> wrote:
> >> Because...?
> >
> >Because any situation where you are multi-homed, if either home fails 
> >and your software can't try both to see which one is working, there's a 
> >good chance it will fail to connect.
> 
> Nice how this thread evolves:
> - Use PI space, it works (but is not really scalable)
> - No, no, get with the program. Get multi-homed on PA.
> - We don't actually have anything for servers and multi-homed PA.
> - Every piece of software has to do HE because of how multi-homed PA fails.
> 
> Great way of advertising IPv6.

Every piece of client software should already support multi-homing.
This is about supporting it *better* so there isn't the big delay
when switching to the second address because the link at the other
end is down.  The stack should be choosing source addresses which
have working outbound paths.  Selecting destination addresses is
up to the application.

This is about making the error cases work better which quite frankly
should have been done years ago for IPv4.  The only ones that don't
benefit from this is GLB developers.

Mark

> _______________________________________________
> v6ops mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/v6ops
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: [phone]                 INTERNET: [e-mail]