MapReduce: Review and open challenges

• Published in: Scientometrics 2016-10, Vol.109 (1), p.389-422
• Main Authors: Hashem, Ibrahim Abaker Targio, Anuar, Nor Badrul, Gani, Abdullah, Yaqoob, Ibrar, Xia, Feng, Khan, Samee Ullah
• Format: Article
• Language: English
• Subjects: Authoring; Bibliometrics; Big Data; Computation; Computer applications; Computer Science; Data management; Data processing; Documents; Exploration; Fault tolerance; Information Storage and Retrieval; Library Science; Reviews; Scientific visualization; Visualization
• ISSN: 0138-9130; 1588-2861
• DOI: 10.1007/s11192-016-1945-y

The continuous increase in computational capacity over the past years has produced an overwhelming flow of data or big data, which exceeds the capabilities of conventional processing tools. Big data signify a new era in data exploration and utilization. The MapReduce computational paradigm is a major enabler for underlying numerous big data platforms. MapReduce is a popular tool for the distributed and scalable processing of big data. It is increasingly being used in different applications primarily because of its important features, including scalability, fault tolerance, ease of programming, and flexibility. Thus, bibliometric analysis and review was conducted to evaluate the trend of MapReduce research assessment publications indexed in Scopus from 2006 to 2015. This trend includes the use of the MapReduce framework for big data processing and its development. The study analyzed the distribution of published articles, countries, authors, keywords, and authorship pattern. For data visualization, VOSviewer program was used to produce distance- and graph-based maps. The top 10 most cited articles were also identified based on the citation count of publications. The study utilized productivity measures, domain visualization techniques and co-word to explore papers related to MapReduce in the field of big data. Moreover, the study discussed the most influential articles contributed to the improvements in MapReduce and reviewed the corresponding solutions. Finally, it presented several open challenges on big data processing with MapReduce as future research directions.