A Survey of Speculative Execution Strategy in MapReduce

Liu, Qi; Jin, Dandan; Liu, Xiaodong; Linge, Nigel

Abstract

MapReduce is a parallel computing programming model designed to process large-scale data. Therefore, the accuracy and efficiency for computing are needed to be assured and speculative execution is an efficient method for calculation of fault tolerance. It reaches the goals of shortening the execution time and increasing the cluster throughput through selecting slow tasks and speculative copy these tasks on a fast machine to be executed. Hadoop naïve speculative execution strategy assumes that the cluster is homogeneous, and this assumption leads to the poor performance in heterogeneous environment. Several speculative execution strategies which aim to improve the MapReduce Performance in the heterogeneous environments are reviewed in this paper like LATE, MCP, ex-MCP and ERUL, then the comparison between these methods are listed.

Citation

Liu, Q., Jin, D., Liu, X., & Linge, N. (2016, July). A Survey of Speculative Execution Strategy in MapReduce. Presented at the 2nd International Conference on Cloud Computing and Security; Lecture Notes in Computer Science, Nanjing, China

• Presentation Conference Type: Conference Paper (published)
• Conference Name: the 2nd International Conference on Cloud Computing and Security; Lecture Notes in Computer Science
• Start Date: Jul 29, 2016
• End Date: Jul 31, 2016
• Acceptance Date: May 1, 2016
• Online Publication Date: Nov 1, 2016
• Publication Date: 2016
• Deposit Date: Jan 30, 2017
• Electronic ISSN: 1611-3349
• Publisher: Springer
• Pages: 296-307
• Series Title: Lecture Notes in Computer Science
• Series ISSN: 0302-9743
• Book Title: Cloud Computing and Security
• ISBN: 978-3-319-48670-3; 978-3-319-48671-0
• DOI: https://doi.org/10.1007/978-3-319-48671-0_27
• Keywords: Hadoop, Map Reduce, Speculative execution, Heterogeneous environment
• Public URL: http://researchrepository.napier.ac.uk/Output/451268