WebMapReduce: An Accessible and Adaptable Tool for teaching Map-Reduce Computing

摘要

WebMapReduce (WMR) is a strategically simplified user interface for the Hadoop implementation of the map-reduce model for distributed computing on clusters, designed so that novice programmers in an introductory CS courses can perform authentic data-intensive scalable computations using the programming language they are learning in their course. WMR currently supports Java, C++, Python, and Scheme computations, and can readily be extended to support additional programming languages, and configured to adapt to the practices at a particular institution for teaching introductory programming. The open-source system is designed to give beginning CS students experience with parallel computing and exposure to concepts of parallelism, at a wide variety of institutions with diverse curricular choices and cluster resources. Potential applications in courses at all undergraduate levels are indicated, and implementation of the WMR software is described.