Transactions on Shared Peer to Peer Dataspace

Abstract

Applications of peer to peer distributed algorithms are very popular in these days. Main reason for this is highly dynamic nature of p2p systems and can achieve bet- ter performance than traditional client server computing using p2p systems. Data used in these distributed algorithms are stored in local storage of each node with enforcing rules and this will reduce the overall performance of distributed appli- cations. With respect to temporal and spatial decoupling concept, coordination model like shared persistent dataspace has proven to be well adopted and suitable for distributed p2p system. Through this research we are trying to combine the "Shared Persistent Dataspaces" paradigm and prominent features of p2p communi- cation model to create shared distributed memory model to perform transactional operations. Using this model we are able to address the issues in data management and to provide robust and easily manageable interface for programmer to manage data in distributed environments. Implementation wise our aim is to create decen- tralized p2p shared dataspace implementation on top of JXTA peer to peer overlay. Local storage of each node contributes to the space of the coordination model. Tu- ple location is not critical in this model. All the dataspace operations (In, Out, Take) are handle through by communication model. In the beginning, model is create to perform basic dataspace operations and as a second step model is extend to perform transactional operations. Transactions are identi ed as set of bundled In, Out, Take operations. Each node in the overlay can perform its own set of transactional operation. Communication model and data handling operations will ensure the correctness and validity of each transactional operation. With compar- ing other dataspace transactional implementation this model introduce novel way of handling transaction without main deployment and transaction handler. Nodes can perform its own transaction without considering infrastructural complexity. We describe our approach, along with the design of a prototype implementation and its quantitative evaluation with this report.