MapReduce System over Heterogeneous Mobile Devices

Authors:
Peter R. Elespuru;Sagun Shakya;Shivakant Mishra
Affiliations:
Department of Computer Science, University of Colorado, USA 80309-0430;Department of Computer Science, University of Colorado, USA 80309-0430;Department of Computer Science, University of Colorado, USA 80309-0430
Venue:
SEUS '09 Proceedings of the 7th IFIP WG 10.2 International Workshop on Software Technologies for Embedded and Ubiquitous Systems
Year:
2009

Citing 6
Cited 4

Scans as Primitive Parallel Operations

IEEE Transactions on Computers
Parallel Prefix Computation

Journal of the ACM (JACM)
Web Search for a Planet: The Google Cluster Architecture

IEEE Micro
Robust System Design with Built-In Soft-Error Resilience

Computer
Introduction to Information Retrieval

Introduction to Information Retrieval
Improving MapReduce performance in heterogeneous environments

OSDI'08 Proceedings of the 8th USENIX conference on Operating systems design and implementation

Misco: a MapReduce framework for mobile systems

Proceedings of the 3rd International Conference on PErvasive Technologies Related to Assistive Environments
Scheduling for real-time mobile MapReduce systems

Proceedings of the 5th ACM international conference on Distributed event-based system
Computing while charging: building a distributed computing infrastructure using smartphones

Proceedings of the 8th international conference on Emerging networking experiments and technologies
HAT: history-based auto-tuning MapReduce in heterogeneous environments

The Journal of Supercomputing

Quantified Score

Hi-index	0.00

Visualization

Abstract

MapReduce is a distributed processing algorithm which breaks up large problem sets into small pieces, such that a large cluster of computers can work on those small pieces in an efficient, timely manner. MapReduce was created and popularized by Google, and is widely used as a means of processing large amounts of textual data for the purpose of indexing it for search later on. This paper examines the feasibility of using smart mobile devices in a MapReduce system by exploring several areas, including quantifying the contribution they make to computation throughput, end-user participation, power consumption, and security. The proposed MapReduce System over Heterogeneous Mobile Devices consists of three key components: a server component that coordinates and aggregates results, a mobile device client for iPhone, and a traditional client for reference and to obtain baseline data. A prototypical research implementation demonstrates that it is indeed feasible to leverage smart mobile devices in heterogeneous MapReduce systems, provided certain conditions are understood and accepted. MapReduce systems could see sizable gains of processing throughput by incorporating as many mobile devices as possible in such a heterogeneous environment. Considering the massive number of such devices available and in active use today, this is a reasonably attainable goal and represents an exciting area of study. This paper introduces relevant background material, discusses related work, describes the proposed system, explains obtained results, and finally, discusses topics for further research in this area.