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



David Conrad wrote:
[..]
It's a bit sad that we didn't take the opportunity to define a new API that would return 'network handles'

SCTP & HIP. Please look at these.

IP is just for sending packets from A to B.

TCP was (afaik, as I am far from old enough ;) created just to send packets to two hosts as long as they can talk to each other. UDP the same. At that time nobody renumbered as as one just got a /8 and expected never to do this.

SCTP was created to avoid all those problems during renumbering as the hosts can notify each other where they are going to (and other cool features).

HIP was created just to avoid the whole debacle completely from the start, using HIP's instead of IP's.


Now the big problem is, how do you get programmers to start adding SCTP and/or HIP support to all their programs... Apache/IIS/Lighttpd don't support SCTP, and Firefox/Safari/IE don't support it either. That is the first step to take. But like IPv6, they are adding something new, and that might cause issues and delays when trying to use those protocols, thus they most likely rather not. Also, especially in the case of HTTP, the connections are shortlived and load-balancer work fine already with TCP/HTTP, those and a lot of other gear also needs to start supporting SCTP/HTTP then all of a sudden. Like IPv6 this is another nice chicken/egg issue. One advantage here is that you just have to touch the endhosts and not the routing infrastructure.


Oh and indeed before you ask SCTP and/or HIP solve a part of the multihoming riddle too. They just don't solve the "I need this traffic to go this fast and there and here" and other such tricks.

Greets,
 Jeroen

Attachment: signature.asc
Description: OpenPGP digital signature