|
Comcute is a distributed computing system created by scientists from , Faculty of Electronics, Telecommunications and Informatics. It’s a first polish volunteer computing project that has been made as alternative to BOINC system. The purpose of such systems is to solve complex scientific problems and to provide computing power with low cost. Moreover, a main goal of the Comcute is to provide continuous activity of processors in case of crisis like the explosion in Fukushima Nuclear Power Plant. In 2010 the Comcute system has got financial aid from polish Ministry of Science and Higher Education in the field of C4ISR under the grant number OR00010811. Comcute features Comcute works with any modern internet browser - you don’t need to install any additional software to start. Furthermore, it can be characterized by following features: * Free and anonymous access; * Participation is safe; * At any moment, without any consequences, you can resign from being a volunteer; * You can participate in solving real scientific problems. System architecture The three layered architecture has been elaborated for the Comcute system. A layer W is responsible for task and data distribution, as well as for delivery of execution modules and data packages for processing. A layer S organizes the processing, in a layer I the processing is being generically carried out. Passing of results starts at the layer I and goes up to the layer W through the layer S. Data packages distribution for tasks Task distribution in the Comcute system is not limited by the system itself, and it’s merely defined by types (classes) of computation applications and their own inherent characteristics. A Comcute user should only deliver task code modules with relevant data packages and may control the distribution processes. The Comcute system does not confine the range of processing models, and allows running of almost any of currently used distributed network processing paradigms, e.g. cloud or grid computing. Implementation specifications In the implementation of the Comcute system, several classes of functionalities have been realized. Among others, one may enumerate the operational functionalities and software components like: user interface, inter-node communicator, the layer S interface, workload monitor, task status monitor, task execution controller, repository for tasks, task distributor, results collector, repository of results, and many others. They are characterised below: User interface * reception of high-level applications for execution; * controlling of tasks execution status; * sending results to a user. Inter-node communicator * general communication with servers of the layer W; * communication with subsets of servers in the layer W; * sending messages to selected server sub-groups; * reception of messages form selected servers of the layer W. Layer S interface * reception of messages form selected servers of the layer S; * data sending to selected servers S; * reception of results from servers of the layer S. Workload monitor * determination of inner workload of a computing nodes; * global determination of workload of servers in layers W and S; * generation of ordered list of workloads of servers in a layer. Task status monitor * status control of tasks being in execution; * reporting of task execution statuses. Task execution controller * task partitioning into execution module; * distribution of modules with task codes; * reception and analysis of partial results; * verification of results of task execution; * verification of task termination conditions; * consolidation of partial results into complete ones. Repository for tasks * storing and delivering of execution codes for tasks; * delivering and storing of data packages for tasks; * collecting and storing partial results of tasks. Task distributor * deploying of tasks on servers in the layer S; * distribution of data packages for processing; * reception of results form execution servers in layer S. Results collector * starting of execution code for results consolidation; * storing of results in results repository. Repository of results * storing of complete final results; * results presentation for Comcute users; * verification of execution of the processed tasks. Selected usage * Genetic Positioning of Fire Stations * Image Processing Techniques * Mersenne Number Finding * Distributed Detection of Selected Features in Data Streams
|
|
|