APNIC Home APNIC Home
Info & FAQ |  Resource services |  Training |  Meetings |  Membership |  Documents |  Whois & Search |  Internet community

You're here:  Home  Mailing Lists global-v6 


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [GLOBAL-V6] IPV4 to IPv6 migration



Hi,

On Tue, Jun 03, 2008 at 07:27:45AM -0700, David Conrad wrote:
> Since applications cache IP addresses (be they IPv4 or IPv6), can you  
> point to _application_ code that copes with multiple addresses in  
> parallel and more specifically, transitioning between one address and  
> another?

Most applications that I'm aware of cache *destination* addresses, not
*source* addresses.  The "a machine can cope with multiple addresses"
statement is talking about source addresses (which, today, usually works
with IPv4 as well, but many stacks seem to have a "preferred" v4 address
plus "aliases", while with IPv6, usually they are "just addresses")

Caching destination addresses for "short" periods of time isn't going
to harm the "add new prefix, wait few days, put new prefix into DNS,
wait few days, deprecate old prefix, ..." process.

Caching destination addresses for unlimited time is *bad*.

> E.g., how do you tell an application to 'deprecate' one address over  
> another?

Applications usually don't explicitely specify *source* addresses.  The 
OS picks the right source address, and the OS knows about depreciation.

If the application insists on specifying the source address for connections,
it needs a way to figure out if something has changed - which, for example, 
bind seems to do quite well (as an example of UDP based server applications 
that need to make sure that the source address of the return packet matches 
whatever address the client sent his packet to).

If something is hardwired in a config file, this falls under the 
"good planing makes renumbering less painful" category.


I'm not really sure what point you're trying to make, though...?

(Note that I never said "IPv6 makes renumbering automatic", I just think
that the impact can be lots less disruptive with IPv6 renumbering than
with IPv4 renumbering - and I've done both a number of times, and seen
the difference)

Gert Doering
        -- NetMaster
-- 
Total number of prefixes smaller than registry allocations:  110584

SpaceNet AG                        Vorstand: Sebastian v. Bomhard
Joseph-Dollinger-Bogen 14          Aufsichtsratsvors.: A. Grundner-Culemann
D-80807 Muenchen                   HRB: 136055 (AG Muenchen)
Tel: +49 (89) 32356-444            USt-IdNr.: DE813185279

Attachment: pgpOWHvCFr79n.pgp
Description: PGP signature