This paper describes the various routing geometries deployed by recent
DHT proposals. These routing algorithms are resilient to random node
failures, a major characteristic of P2P systems. These routing
algorithms are also aware of proximity of nodes to reduce latency.
In my opinion, the time consuming part in a P2P system is the transfer
of actual data. DHT routing algorithms is merely there to help looking
up the location of data from a key and the time it takes to complete its
job is small comparing to the actual data transfer. So, I don't believe
performance, especially in terms of latency, is very important for these
algorithm. I'd care about static resilience, simply because P2P systems
are useless without knowing where the data is. Therefore, I only find
the first part of the paper interesting.
Received on Mon Nov 21 2005 - 11:03:45 EST
This archive was generated by hypermail 2.2.0 : Mon Nov 21 2005 - 11:03:46 EST