Enabling Data Processing under Erasure Coding in the Fog
Published in 48th International Conference on Parallel Processing, 2019
Recommended citation: Jad Darrous, Shadi Ibrahim. "Enabling Data Processing under Erasure Coding in the Fog". In the 48th International Conference on Parallel Processing (ICPP 19), Aug. 2019, Kyoto, Japan.
Abstract - In recent years, Fog computing has been widely adopted as an extension to clouds by placing cloud services close to the users. While Fog promises to provide low-latency by enabling computation close to the data sources, it imposes several challenges for Big Data analytics including limited storage capacity, and computation and network heterogeneity. In this work, we argue that erasure coding (EC) can be applied as an alternative to replication to store data in Fog environments. Moreover, by making map task scheduling network-aware, we might achieve similar performance as replication but with half storage cost.