Review - Serving DNS using a Peer-to-Peer Lookup Service

This paper introduces DDNS, a system that explores the idea of using
DHTs to provide DNS service. DDNS is based on P2P Chord architecture and
uses the idea of DNSSEC to sign records to validate their authenticity.
The paper concludes that this architecture is not a good enough
replacement for conventional DNS because it is slower by an order of
magnitude.

The strength of this paper is that it looks into the possibility of
using DHTs for DNS. It outlines the current problems with DNS and kind
of comes up with some arguments for a P2P based DNS in that it
simplifies DNS management and makes the whole system more resilient.

However, I believe this paper very poorly explains the design and
implementation. I believe that an example would give a lot more insight
on how they implemented this system, which is currently very confusing -
especially how they do the signing of records. It also leaves a lot of
questions unanswered.
1. How does removal of a record take place in DDNS?
2. Is the system still based on the assumption that conventional DNS
makes, i.e. DNS records change infrequently?
3. How to propagate changes to records to all nodes that contain the record?
4. They propose powerful servers to have multiple node in Chord. As we
discussed last time, this is a bad idea, especially when the powerful
node fails.

Hierarchy in DNS works great because it distributes the load from the
root to the top level domains (TLD) nameservers and so on. Maybe some
hierarchy in DDNS might help reduce the number of RPCs.
Received on Wed Nov 09 2005 - 20:39:55 EST