Tutorial #9

1. In the RR examples given in the lecture, the TTL field is set to 86400. What is the significance of this strange number?

2. The DNS is described as a "distributed database" of RRs.
   1. What does this mean?
   2. What is the alternative, and why is it generally regarded as unworkable? (Optional philosophical discussion question: How does this second alternative compare (conceptually) with the various Web "search engines" such as Google and AltaVista?)

3. A nameserver acts not only as a server, but also as a client under certain circumstances. What are these circumstances?

4. Why^[1] should each nameserver know the IP address of its parent instead of its domain name? Similarly -- when configuring an Internet-connected computer, why is the nameserver always specified as an IP address, not as a domain name?

5. Nameservers are usually (always?) configured to know the IP address of at least one root nameserver, as well as that of their parent nameserver. Why is this?

6. Why do you suppose the rules for nameservers in the Internet are so stringent in the matter of off-site "replication" servers?

7. What is a reverse lookup in the DNS, and why is it regarded as a significantly harder problem than normal lookups?

8. What is the significance of the fact that machine luga.latrobe.edu.au appears in an MX RR (Resource Record) for machine ironbark.bendigo.latrobe.edu.au? What facility has to be enabled on luga.latrobe.edu.au for this to work?

9. A nameserver query contains a parameter bit which is set to 1 if recursion is desired at the server and 0 otherwise. What would you expect to be the result of queries in each of these situations?

10. Research & discussion question: Most (all?) implementations of the domain name system allow abbreviations of names so that, for example, the name ironbark resolves to a correct address for machines co-located at the Bendigo campus. How is this handled by the DNS, and whereabouts is it implemented (ie, in the nameserver/s or in the resolvers)? What about ironbark.bendigo -- can this be handled?

11. Implementation question^[2]: The standard suggests that when a program needs to find the domain name associated with an IP address, it should send an inverse query to the local server first and domain in-addr.arpa only if that fails. Why?