¹School of Software, Central South University, Changsha, P.R China
²School of Information Science and Engineering, Central South University, Changsha, P.R China

---

Abstract—The combination of Data warehousing and data analysis techniques such as OLAP (Online Analytic Processing) and data mining through the Hadoop framework is an innovative way to treat large volumes of data. However, this way poses serious scheduling and combining tasks issues that bring more challenges. In this paper, we propose strategies to answer these questions, namely parallel OLAM (Online Analytic Mining) MapReduce Operators and a Reactive Scheduling Policy. OLAM MapReduce Operators divide jobs into two parts, the first includes all the operators that are used to create an OLAM CUBE and the second includes those who exploit the cube by data mining algorithms. The proposed policy coordinates the workflow generated by these operators, relying on model-based events. Our simulation experience shows that our strategy has a cumulative force that it reduces the execution time of the entire cluster at each request.

Index Terms—OLAP, data mining, parallel OLAM, MapReduce Operators, Reactive Scheduling Policy

[PDF]

Cite: Shengbing Ren, Dieudonne Muheto, "A Reactive Scheduling Strategy Applied On MapReduce OLAM Operators System," Journal of Software vol. 7, no. 11, pp. 2649-2656, 2012.