A Privacy Data-Oriented Hierarchical MapReduce Programming Model

Abstract

To realize privacy data protection efficiently in hybrid cloud service, a hierarchical control architecture based multi-cluster MapReduce programming model (the Hierarchical MapReduce Model,HMR) is presented. Under this hierarchical control architecture,  data isolation and placement among private cloud and public clouds according to the data privacy characteristic is implemented by the control center in private cloud.  And then, to perform the corresponding distributed parallel computation correctly under the multi-clusters mode that is different to the conventional single-cluster mode, the Map-Reduce-GlobalReduce three stage scheduling process is designed. Limiting the computation about privacy data in private cloud while outsourcing the computation about non-privacy data to public clouds as much as possible, HMR reaches the performance of both security and low cost.

DOI: http://dx.doi.org/10.11591/telkomnika.v11i8.3083