konspire2b is a new peer-to-peer file sharing system that is quite different from anything else out there. We took the common paradigm and reversed it. This paradigm shift lead to two interesting features:
- a pleasantly different user interaction model
- near-optimal data distribution characteristics (both in terms of latency and load balancing)
k2b adds digital signatures to this new model to achieve semi-anonymous accountability and trust (i.e., the death of spam and RIAA decoy tactics). A public beta test will be starting soon. You can sign up to join this test at: http://konspire.sourceforge.net
k2b compiles as native code (written in C++) for the following platforms:
- Windows 95/98/NT/2000/XP
- GNU/Linux Intel (X86)
- GNU/Linux PPC
- FreeBSD Intel (X86)
- Mac OS X
Read on for more details.
The common paradigm in P2P (and in the majority of other internet systems) can best be described as a source-discovery model. Resource sinks are active, while resource sources are passive. The sinks seek out sources that have content they need. Gnutella, Freenet, eDonkey, FastTrack—even the Web itself, augmented with Google—all are source-discovery systems. All of these systems have scalability problems: in general, the sinks for a popular piece of data far outnumber the sources for that data, especially when that piece of data is first made available. They also share a common user interaction primitive: the search (better described as “search wait comb re-search wait comb re-search groan”).
In the end, data is transmitted from source to sink. Source discovery is just the means by which the transmission is orchestrated. The point to note is that source discovery is not the only way to initiate data transmissions.
Reversing the common paradigm, we come up with what can best be described as a sink-discovery model. Sinks are passive, and the sources seek out sinks that want the content they have. Since the sources are driving the interaction, they can orchestrate a more efficient data distribution tree. In fact, with no extra network analysis or computation, near-optimal distribution trees can simply emerge in a sink-discovery system. In addition, the interactive search—at least as we know it—is no longer possible.
With this new paradigm, many new user-interaction models are possible (including non-interactive search). We chose a channel-based subscription model, and we integrated digital signatures for accountability and trust. This mode of user interaction eliminates the frustration of searching while still supporting the transmission of novel (e.g., zero-day), high-quality content.
Since no existing P2P system uses sink discovery, we built konspire2b as a pure sink-discovery system. There is no support for “falling back” to a source-discovery operation mode (in other words, k2b simply does not support searching). With a pure sink-discovery system like k2b, we can evaluate this new model against the existing source-discovery model in terms of both performance and usability.
konspire2b has been in active development for 11 months and has gone through nearly a dozen rounds of internal alpha testing. No known bugs exist—this beta test aims to discover any bugs that our alpha testing has missed.
Related Posts
- Looking for 50 Beta Testers for the Philips Streamium MC-i250
- project fasttrack asking for more beta testers
- LimeWire 2.8.0 BETA
- You say you want a revolution? Meet Kast.
- Shareaza 1.7.0.9 Beta
Loading ...
