Blognet

Blognet is a P2P network based on the existing weblog infrastructure, RSS 1.0, mod_subscription, mod_link, Web Services Description Language ( WSDL) and Simple Object Access Protocol (SOAP).

What is blognet trying to accomplish?

Blognet is trying to accomplish the following goals:

1. Provide a semantic description of weblogs interconnections with the mod_subscription format in order to enable weblog and node discovery and other compelling uses of P2P networks. 2. Facilitate the possibility of WSDL and SOAP service discovery over RSS. With the current RSS specifications this is impossible. 3. Provide RSS users with the ability to link to additional RDF and XML files (other formats are also supported). 4. Accomplish all of this with very few (if any) changes required from RSS producers( Blogger, Moveable Type, etc)

What are the long term plans for blognet?

Blognet provides the base infrastructure for a lightweight P2P network that supports node linkage and ad hoc service discovery. This should facilitate the deployment of complex search algorithms and reputation networks.

It is hoped that bootstrapping blognet today with RSS 1.0 and mod_subscription will provide the basis of a network that will last for a long time and provide a fertile environment for the development of more complex P2P scenarios.

How is this a P2P network?

Blognet allows users to include subscription information within the RSS channel for their blog. Each RSS channel represents a node and subscriptions represent edges within the network. This allows us to build a P2P network by connecting each blog (represented by its RSS channel). WSDL and SOAP provide services for each blog such as searching, reputation, etc.

What differentiates blognet from existing P2P networks?

Blognet incorporates and extends a lot of existing technology into a higher level P2P framework:

* Very thin and simple to develop and work with. Nodes are represented as RSS channels and XML which a lot of developers understand and appreciate. There is also a lot of development within this area outside of blognet which other P2P networks do not benefit from (rss-dev, W3C, etc) * Supports any transport mechanism including HTTP, BEEP, JXTA, Gnutella, etc. Blognet is not a transport mechanism. * Supports a flexible and ad hoc service deployment mechanism based on WSDL and SOAP. This allows developers to incorporate technology from a number of providers into a larger network. * Supports all existing weblog clients assuming they support RSS export. * Blognet is based on RDF and supports very complex metadata scenarios allowing developers to develop very compelling applications. * Blognet is easily extensible via the use of RSS 1.0 modules and or the addition of new services via WSDL service descriptions. The network can be updated without having to update software on every peer in the network.

What software is required to run blognet? What changes are required from RSS publishers?

All that is required is an RSS producer that produces RSS 1.0 with the ability to link to additional RSS files with the mod_link format.

The links can be defined internally with a GUI or with a interactive API and a 3rd party application.

We are also developing a reference implementation of a blognet client that supports Windows, OSX, Linux and BSD that supports RSS subscription, and deployment of services within the users RSS feed. This acts as an RSS producer so that the user can publish new articles as well as an RSS reader so that the user can run searches and browse other weblogs.

Blognet also supports the use of existing blogging software (Radio, Moveable Type, etc) through the use of an integration system that produces RSS 1.0 compatible with our needs.

Can I use blognet to distribute MP3 files? What copyright issues does blognet face?

Blognet was not designed for the distribution MP3 files. You may be held liable for any content you distribute.

In truth there are very few (new) Intellectual Property issues with blognet as it was designed for news and if there are IP issues the person hosting the node is somewhat identifiable. This is both good and bad. Either way we have this situation with current weblog (and Internet) environment.

How is node discovery accomplished with blognet

The entire network can be walked just like HTTP robots walk the current Internet. This is accomplished by starting at a few root RSS channels that have exposed subscription edges and spanning outward.

Search algorithms can use this code and other advanced techniques (TFIDF, reputation, indices, semantics) to find resources within the network.

How bad is node availability and network churn within blognet?

Blognet nodes (represented as RSS channels) offload their services to 'agents' within the network. These services can run on any type of network including a convention website with a SOAP webservice or a JXTA (P2P) peer running SOAP. The only requirement is that the services is exposed through WSDL.

This allows us to support very complex network configurations. Users can publish from a node with low availability (a laptop on a 24kbps modem) and use a service provided by a high availability agent running on a HTTP/SOAP service on a high bandwidth T1 line.

It is also possible to run blognet with a more chaotic P2P infrastructure such as JXTA in order to support services such as chat, real time event delivery, etc.

How is search accomplished within blognet?

Initially blognet will use a search algorithm similar to Gnutella (spanning all nodes forward). This decision was made because:

1. We wanted to have blognet available for users sooner rather than later. 2. It is simple and something to fall back upon. 3. The blognet infrastructure will easily support innovation within this area without breaking any existing search infrastructure. 4. We want to work with other search technologies such as Neurogrid which may take a long time to incorporate into blognet.

How does blognet relate to Freenet, JXTA, Gnutella, Alpine, Chord, etc?

Blognet is only a metadata network on top of existing networks. It is somewhat virtual in that we do not care what type of network you are using for transport of data. Blognet users can use JXTA, Chord, HTTP, BEEP, etc for their network transport and blognet can operate just fine.

The only exception would be the use of a transport protocol which a peer does not support. If a blognet peer does not support Chord when this section of the blognet is unavailable.

Why was RSS chosen to represent blognet nodes?

RSS is very compelling because a number of existing technologies support it (Radio, Moveable Type, etc). The RSS 1.0 module extension mechanism allows us to develop on top of the base RSS specification and add additional metadata.

RSS also represents an RDF graph which is updated constantly with new data. This amount of structured information provides a number of very compelling opportunities for application developers.

The mod_link specification allows us to link to agents that provides the services that blognet needs to operate.

How is NAT handled within blognet?

NAT is an issue that transport providers must handle. HTTP/SOAP agents must be present outside of firewalls and must be publicly addressable. If the current node is using a P2P technology it may support NAT bypass.

What are the requirements for becoming a blognet node?

1. Publish an RSS 1.0 channel with at least 1 subscription. 2. Deploy a WSDL service with mod_link that supports Reptile search.

Ideally the node would support a number of subscriptions (to help discovery). Optional services can be deployed in order to provide more functionality to other blognet peers.