QoS-aware proactive data replication for big data analytics in edge clouds

We are in the era of big data and cloud computing, large quantity of computing resource is desperately needed to detect invaluable information hidden in the coarse big data through query evaluation. Users demand big data analytic services with various Quality of Service (QoS) requirements. However, cloud computing is facing new challenges in meeting stringent QoS requirements of users due to the remoteness from its users. Edge computing has emerged as a new paradigm to address such shortcomings by bringing cloud services to the edge of the operation network in proximity of users for performance improvement. To satisfy the QoS requirements of users for big data analytics in edge computing, the data replication and placement problem must be properly dealt with such that user requests can be efficiently and promptly responded. In this paper, we consider data replication and placement for big data analytic query evaluation. We first cast a novel proactive data replication and placement problem of big data analytics in a two-tier edge cloud environment, we then devise an approximation algorithm with an approximation ratio for it, we finally evaluate the proposed algorithm against existing benchmarks, using both simulation and experiment in a testbed based on real datasets, the evaluation results show that the proposed algorithm is promising.