Decentralized Distributed Processing Michael Tao
Abstract Increasing efficiency Power of independent processing limited Distributed Processing By decentralizing, anyone can join the network to aid others
Background Projects with large magnitude of data to analyze such as Seti already use applications such as Boinc Currently there is no way for 3rd parties to join this network without publicizing itself and having each node manually add the project to the list
Purpose Communicate and inject jobs to peers in queues Increasing efficiency of data analysis with the aid of multiple computers which aren't necessarily available
Architecture Local listener for new tasks waits to release work to the network to accomplish some task Nodes on the network which receive the new work will initialize clients, with variable number of clients at any one time
Developments Computer chatting / file transfers Computer load deciding Modular analysis tool
Results Chatting between two computers Individual clients can now do tasks given to them passed from the main upon initialization