Predictive Load Balancing for Data Mining in Distributed Systems

Provided by: Journal of Theoretical and Applied Information Technology
Topic: Data Management
Format: PDF
Load Balancing is the key attribute in distributed systems to ensure fast processing and optimal utilization of hardware components. The processing elements of varying hardware architecture and computing capabilities in a shared network system constitute the distributed system. In such a heterogeneous environment, load balancing and scheduling become complex and complicated. Dynamic feedback based techniques, predicting the processing time of a task in a node, will be a promising solution to resolve variations of this ubiquitous system. A collaborative agent system comprising mobile and stationary agents captures the feedback from the contributing nodes and the prediction for the proceeding time epoch is made by minimizing the gap between the predicted processing power and the actual processing power of the node in the preceding epoch.

Find By Topic