O Plameras wrote:
Ben Leslie wrote:
On Thu Dec 07, 2006 at 16:52:21 +1100, O Plameras wrote:
*snip*
I have first, second, and third editions. I have the third edition in
front of me.
The book covers the technical process. Unfortunately, it does not
cover the bureaucratic
processes. The processes not covered by the book is the one that I am
revealing to you.
For example, technically aarnet.edu.au can propagate up to the ROOT
Servers.
Technically aarnet.edu.au doesn't need to propagate anything "up" to
the ROOT servers. That is not the way DNS works, rather the client
contacts the ROOT servers and then goes down from there (ignoring any
caching).
So really, aarnet.edu.au doesn't need to propagate anything at all.
Do you mean once aarnet.edu.au enters www.example.aarnet.edu.au IN A
203.7.132.1
it will be propagated ? This is wrong. aarnet.edu.au is only a branch in
the DNS trees.
What does aarnet.edu.au has to do to propagate ? If you can answer this
last question
correctly then we can proceed with the discussion.
How does a query propagate? Well, to paraphrase Section 2.6.2 of that
book that's right in front of you(*):
- The local nameserver gets a request for www.example.aarnet.edu.au. It
doesn't know where this is, but it does know where all the root
nameservers are. So it picks one of those and asks it.
- The root nameserver says, no, I have no idea where
www.example.aarnet.edu.au is. But I do have this list of nameservers
that are authoritative for the .au domain, maybe one of them can help
you. The local nameserver picks one, and sends the query for
www.example.aarnet.edu.au to it.
- The .au nameserver says, no, I have no idea where
www.example.aarnet.edu.au is. But I do have this list of nameservers
that are authoritative for the edu.au domain, maybe one of them can help
you. The local nameserver picks one, and sends the query for
www.example.aarnet.edu.au to it.
- The edu.au nameserver says, no, I have no idea where
www.example.aarnet.edu.au is. But I do have this list of nameservers
that are authoritative for the aarnet.edu.au domain, maybe one of them
can help you. The local nameserver picks one, and sends the query for
www.example.aarnet.edu.au to it.
Do you see where this is going? The query keeps propagating down
different levels until it finally hits a server who says "Ooo! I *know*
this one!" and replies.
It sounds like a lot of traffic, but the local nameserver also caches
all of those replies it got along the way to resolving that query. So if
the next query it gets is for, say, www.monash.edu.au, the local
nameserver will say "I don't know where that is, but I've already got
this list of .edu.au nameservers that is still fresh in my cache, I'll
ask one of those".
In a perfect world, the root servers wouldn't get that much traffic at
all, really, thanks to caching. But there's a *lot* of poorly configured
nameservers out there. http://www.bind9.net/dnshealth/ makes the claim
that 98% of queries to the root servers are unnecessary, and is full of
lots of other interesting DNS-related factoids.
Anyway, the point of all of this is, *all* of the propagation in DNS
happens downwards. The root nameservers seriously don't know the first
thing about subdomains of aarnet.edu.au, and you can verify this by
sending a non-recursive query to one (using the +norecurse option to
dig, for example). Not only do they not know whether a particular
subdomain exists, they really don't care whether the object it resolves
to (not all DNS records are IP addresses...) belongs to the domain
owner, or Dexter Plameras, or whether it even exists on the public
Internet at all.
I desperately hope this clears up any misunderstandings.
* I have the fourth edition. May be slightly different for others.
Cheers,
--
Pete
--
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
