[dns-operations] Introducing CNAME Flattening: RFC-Compliant CNAMEs at a Domain's Root

[Mark Andrews]

In message <21310.50033.650900.982948 at tale.kendall.corp.akamai.com>, David C Law
rence writes:
> Stephane Bortzmeyer writes:
> > I was not able to find a real example: all the companies mentioned in
> > the article as being happy users have their Web server at CloudFlare,
> > where this service is useless.
> 
> I agree, it is a little peculiar, with their anycast HTTP servers.
> You can see that some of these sites have www names that use a CNAME
> intermediary, and conceivably they have configured their apex to use
> that same CNAME with CloudFlare's flattening feature.  They're also
> all giving me the same result from clients in both New England and
> from Tokyo, so it does make it difficult to discern from the outside
> that anything special is going on with an intermediary name.
> 
> Mark Andrews writes:
> > 	Or one can add SRV or some other record that does the name
> > 	to server mapping and not have to do all this behind the
> > 	scenes stuff.
> 
> Lack of browser support is a big barrier to adoption here, and browser
> vendors don't seem particularly eager to tackle it.  
> https://bugzilla.mozilla.org/show_bug.cgi?id=14328
> is still in state New, and it was opened in 1999.
 
The big barrier is people saying "there is a big barrier".

FreeBSD is using _http._tcp today for some of their services.

> Me?  I love the vision of it, obviating hacks around the apex CNAME
> restriction, but it just isn't practical right now.  People
> like the econsultancy.com CTO quoted in the article have a problem
> with wanting to brand their sites without www (as you yourself
> indicated was an issue in your draft back in 2002).  Resolving the
> CNAME internally solves their problem and works with existing
> clients.

Which doesn't stop a proper solution being defined which leaves you
ties to a single hoster.
 
> Bring me a world in which SRV for HTTP without location bar
> redirection is commonplace, and we'd be happy to encourage its use.
> 
> Anthony Eden writes:
> > While CloudFlare did not give any credit to previous work done
> > (which sort of pisses me off, but whatever), they are essentially
> > implementing the same thing that Amazon did with their ALIAS
> > implementation, the same thing that we did with the DNSimple ALIAS
> > implementation, and the same thing that DNSMadeEasy did with ANAME
> > records.
> 
> FWIW Akamai has been doing it since 2003, but I don't feel
> particularly put off that they didn't itemize how their competitors
> do similar things.
> _______________________________________________
> dns-operations mailing list
> dns-operations at lists.dns-oarc.net
> https://lists.dns-oarc.net/mailman/listinfo/dns-operations
> dns-jobs mailing list
> https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: marka at isc.org