Question about "proper" way to run v6/v4 website
niels=cluenet at bakker.net
Wed May 2 13:50:26 CEST 2007
* dougb at dougbarton.us (Doug Barton) [Wed 02 May 2007, 08:06 CEST]:
I have little to add to Remi's reply but would like to say the following:
> So I'll ask the question again. What operational goal are you servicing by
> enabling v4-only clients to query a name server that does v6 queries?
Are you suggesting that instead of equipping hosts with dual stacks,
instead we double our hardware expenses and use separate machines for
serving IPv4 and IPv6 networks?
Or in the case of a single application running on a dual-stacked
machine, double our memory usage by effectively running one instance
for v4 and one for v6?
>> The same reason applies to answering differently based on source IPv4
> Actually I think you're misapplying the analogy here for at least two
> reasons off the top of my head. The most important reason being that by
> acting deterministically on the basis of receiving a query over v6,
> we're not guessing at anything.
You are guessing, as the machine that sends you the query is unlikely
to be the end user's machine.
> (The other reason is that at least one
> company has made a multi-billion dollar business out of "guessing"
> things about your traffic based on your IP address.)
On your IP address or on the IP address of your resolver? Either way,
you cannot maintain that this is a BCP in any sense of the word on the
current Internet. (There is no more than a strong correlation between
> BTW, there is at least one other problem with what I proposed that no
> one has mentioned yet, what about clients that are v6-capable behind
> a resolving name server that isn't? But I think my main point is still
> valid. WHILE WE ARE IN THIS TRANSITION PERIOD, how can we make
> migration to v6 traffic simpler, and more transparent than it is now?
That wasn't your main point. Your main point was questioning the wisdom
of making applications dual-stacked.
Several posts upthread an idea was put forward to make it easier for end
users to choose whether to connect over v4 or v6 to a service. The
proposed idea was a different hostname. By special-casing DNS answers
based on the transport of the question you do not make it easier for end
users to make such a choice for the reasons outlined also by yourself.
"The Mac doesn't have a one-button mouse, it has a five-button mouse, with
four of the buttons on the keyboard."
-- Peter da Silva <peter at taronga.com>
More information about the ipv6-ops