Application Design on Information Centric Networks

Over a span of three decades the requirements by users and applications of the Internet have changed from what they were in the beginning. It is used more for distribution than resource sharing which is what its architecture was designed for. This has brought new challenges. To solve this, academic research in the field of Future Internet Architectures (FIA) has led to different new Internet Architecture proposals known as Information Centric Network (ICN) architectures. However, little has been done to validate how these architectures fit into the real world. The aim of this thesis is to validate how the PURSUIT architecture, one of the FIA proposals, fits in the real world. This will be done through porting an open source collaborative editor on top of Blackadder, the prototype of the PURSUIT architecture. This architecture follows a publish/subscribe model where data is published into, and subscribed from the network by applications. Unlike Host Centric Networks (HCN) networks, ICN networks do not identify hosts, rather the data is named and any interested party would use the names to access data from the network. Due to this nature of ICN networks, this thesis will propose a server-less design as the approach to developing distributed applications on top of ICN networks, through porting a collaborative editor. The porting experiment will follow this server-less based design as opposed to the more common client/server model that the collaborative editor has been designed in. After the porting exercise, we shall evaluate our findings through measuring quality metrics and performing static analysis. The quality metrics will show that there will be increase in complexity of the design mainly because of having the client applications and the network dissolve the functions that the server performs. A new dimension of concurrency control of state within a distributed network will be realized. In a client/server model, the requests to the server are serialized, hence not having the concurrency control challenge. However, this will not be researched on in this thesis. We shall propose to have this as a future research area