Joao, Joao Damas wrote:

Andrei,

there are 2 issues here, one is how to provide transport and another how to include records in the root zone (this is the one you are referring to but NOT the one this thread started with).

Actually how to include records in the root zone is a third issue. I agree that as long as one maintains a list of statically configured root servers there is no problem with enabling ipv6 for the root servers.

There is no reason not to enable IPv6 transport. It is dependant only on the operator of the machine.

As I said we don't have definite plans regarding this. At the moment our main priority is the anycast implementation for K. If ipv6 can be achieved as a by-product we will probably do it. A few root servers already provide ipv6 access but I guess that in most cases one will fall back to ipv4 anyway to resolve a name.

People will find ways to use the service (the ones who don't already do so with the available servers) if the service is available, so the first step is to enable the service.

Joao

Thanks, Andrei

On Wednesday, Jun 25, 2003, at 09:00 Europe/Amsterdam, Andrei Robachevsky wrote:

...

Gert Doering wrote:

...

Hi, On Tue, Jun 24, 2003 at 04:46:01PM +0200, Andrei Robachevsky wrote:

...

As for the K (being an operator of this server) we don't have definite plans for ipv6 deployment. Also there are different views on how ipv6 access should be provided to the root zone.

Could you explain this a bit more? I'm curious what's going on.

I was referring to the problem with priming, where one can fit only two AAAA records in the priming response provided that not all DNS implementations support EDNS0 at the moment. One proposal I heard of suggests that two selected root servers provide anycast service in ipv6. I thought it was going to be published as an internet draft but cannot find it.

...

Gert Doering -- NetMaster

Andrei