We asked our resident p2p technology expert, Gwendal Simon, to comment on what the recent Skype outage means for the future of peer to peer computing, as it may be interpreted as a sign of weakness. For added context, check our updated entry on P2P Computing.
“”Skype uses a different type of Peer-To-Peer network than most companies — a proprietary form based on SuperNodes. A SuperNode P2P System is one in which you rely on your customers rather than your own servers to handle the majority of your traffic. SuperNodes are just normal computers which get promoted by the Skype software to serve as the traffic cops for their entire network. In theory this is a good idea, but it does have unique vulnerabilities that have been exposed during the past week. Skype, as a company, has no physical or programmatic control over the most vital piece of its product when the network destabilizes for any reason.
Another issue with SuperNode models concerns system recovery after a crash. A SuperNode-based network can only recover as fast as new SuperNodes can be identified.”
Here is the contribution by Gwendal Simon:
“Skype is undoubtedly one of the most famous peer-to-peer systems over Internet. Skype sofwtare is so efficient that many users consider Skype as a casual phone company, although the system is actually based on mechanisms that were initially designed for file sharing systems.
This small start-up leverages on known advantages of such architectures: scalability, self-healing ability, no (or few) required infrastructure…
Users were quite disappointed when the system experienced a major crash during two days in mid-August. Reported in the official Skype blog , this bug emphasized the deficiency of Skype’s self-healing property.
Many experts have debated about this crash, however very few facts have been made public by the company which uses a close proprietary network protocol.
A nice reaction to this crash comes from Alen Peacock. He claims
that the close proprietary model of Skype is the main cause of this crash. He convincingly argues that if Skype were open, many different programs would form an unique Skype system and this system would be far more reliable because one bug in one program could not affect the whole system. In other words, the self-healing property of a peer-to-peer system does not depend only on the decentralized architecture, but should also rely on a full distribution of the system.
We make here the parallel with a previous study on the Gnutella system where the diversity of actors was also shown as a positive characteristic.
In a classic market view, Gwendal Simon showed that the competition between several actors united in the same system is good for the system itself as it benefits from innovations from several actors.
These competitors try to differentiate but they also try to make the system more attractive. This dilemma results, for Gnutella at least, in a positive evolution. As an example, the seminal mail in the Gnutella Developers Forum said : Â« The network effect implies that the more the network grows, the more powerful and valuable it will be. We have more to gain in cooperation than in competition. Lets try to keep this in mind as we have technical discussions. Â»
In both cases (self-healing and innovation), the diversity of actors appears to be positive. The competition between actors reaches a natural collaborative state as actors know that their main value comes from the network. The recent Skype outages emphasize another positive aspect of the diversity generated by open common system.”