Ad Hoc, Self-Supervising Peer-to-Peer Search Networks

Ad Hoc, Self-Supervising Peer-to-Peer Search Networks – nice paper on a proposed architecture for p2p networks, built on primitives which are so simple they’re obviously right. :-) Abstract:

Peer-to-peer search networks are a popular and widely deployed means of searching massively distributed digital object repositories. Unfortunately, as such networks grow, they place an increasingly overwhelming load on some or all of the participating nodes. We examine how to reduce the load on nodes by allowing them to self-organize into a relatively efficient network, and then self-tune to make the network even more efficient. Unlike previously studied architectures, our “ad hoc, self-supervising” networks avoid restrictions on who a node can connect to or what information can be exchanged. This makes the network topology quite flexible and tuneable. Our results indicate that our ad hoc networks are more efficient than popular supernode topologies for several important scenarios.

One Response to “Ad Hoc, Self-Supervising Peer-to-Peer Search Networks”

  1. Luke
    July 30th, 2005 | 12:43 am

    I’m not really an expert on algorithms. You might even say i’m an ignoramus on the subject. However, i’ve had a few drinks, so i’m more than willing to comment on this post.

    In the interests of full disclosure, I think I had this paper sitting on my desk for a year in one of the P2P conference compilations. Not that i’d actually read it ;)

    It’s an interesting paper in theory, but it seems to me that there are fair reasons why it has not been put into practise. Although the authors position there work as a counter-point to “super-node” based P2P networks, from my brief perusal, it seems it could be more appropriately read as a derivative of such networks.

    Whereas super-node based networks consist of just nodes and super-nodes, the system discussed seems to be proposing a network of nodes of varying degrees of superness. This seems reasonable, as internet connections come in all shapes and sizes. It makes sense to tie the degree of involvement of a node to it’s available bandwith precisely.

    However, it is still working on the same principle as super-node based networks. Although it may be slightly more efficient, it does not overcome the limitations of it’s fundamentally undirected and verbose search architecture. The somewhat fickle nature of the network may also effect it’s integrity, and lead to network splits in practise.

    It’s interesting that the authors did not directly compare their offering with any DHT based search networks, merely commenting on them as unproven. That may have been the case to some degree when the paper was written in 2003, but certainly not today. This also seems a bit rich coming from an unimplemented, proposed solution from the realms in academia. I posit that they didn’t do any direct comparisons with DHT based networks because they saw the writing was on the wall.

    DHTs are where it’s at. Although it seems Skype stole my idea before i’d thought of it, i’m gonna use DHTs in writing my P2P IM network.

    Which i’m gonna start just as soon as my beer supply runs out…