A Programming Framework for Incremental Data Distribution in Iterative Applications

Download Now
Provided by: Monash University
Topic: Big Data
Format: PDF
Desktop grids, networks of non-dedicated workstations, and peer-to-peer systems are cost-effective alternatives to high-end clusters and specialized supercomputers. However, due the fluctuating workloads experienced by nodes on such systems, achieving consistent and good application performance is a challenge. One approach to improving application performance is dynamic reconfiguration. Reconfigurable applications involve either dynamic data distribution - altering the data assignment among processes to improve load balance; and/or support for dynamic process joins and departures - adjusting the degree of parallelism in response to changes in availability of resources.
Download Now

Find By Topic