SDPMN: Privacy Preserving MapReduce Network Using SDN

TL;DR

This paper introduces SDPMN, a scalable framework using SDN to isolate and preserve privacy between applications in MapReduce data centers, enhancing security without significant hardware changes.

Contribution

It proposes a novel SDN-based framework for privacy preservation in MapReduce networks and a heuristic algorithm for optimal rule placement within SDN devices.

Findings

Supports more privacy-preserving networks with existing SDN switches

Heuristic algorithm effectively optimizes rule placement

Framework is scalable and compatible with Hadoop networks

Abstract

MapReduce is a popular programming model and an associated implementation for parallel processing big data in the distributed environment. Since large scaled MapReduce data centers usually provide services to many users, it is an essential problem to preserve the privacy between different applications in the same network. In this paper, we propose SDPMN, a framework that using \textit{software defined network} (SDN) to distinguish the network between each application, which is a manageable and scalable method. We design this framework based on the existing SDN structure and Hadoop networks. Since the rule space of each SDN device is limited, we also propose the rule placement optimization for this framework to maximize the hardware supported isolated application networks. We state this problem in a general MapReduce network and design a heuristic algorithm to find the solution. From the simulation based evaluation, with our algorithm, the given network can support more privacy preserving application networks with SDN switches.