可信固态硬盘:大数据安全的新基础

发布时间：2018-01-12 21:31

本文关键词：可信固态硬盘:大数据安全的新基础　出处：《计算机学报》2016年01期 　论文类型：期刊论文

【摘要】：大数据平台,因其数据多、价值高和存储集中的特点,已经成为对攻击者非常有吸引力的目标.因此,大数据安全是一个非常重要的研究课题.然而,当前保障大数据平台(如Hadoop)数据安全的两种常见方法各有不足:(1)访问控制.存在被外部黑客攻破或内部管理员绕过的风险;(2)数据加密.虽然安全性较高,但加密解密海量数据会增加显著开销.为了同时满足大数据应用对数据存储的高安全和高性能要求,文中提出可信固态硬盘(TrustedSSD),它提供安全增强的存储设备接口和协议,使得用户可以对存储中的数据施以细粒度的访问控制,从而保障存储中数据的安全.文中深入分析了可信固态硬盘的安全性,并详细介绍了系统设计与实现中的挑战和应对.实验结果表明,无论是在合成的还是真实的工作负载上,可信固态硬盘的运行开销不到3%.因此,可信固态硬盘有望成为大数据安全的新基础.
[Abstract]:Large data platform, because of its many data, high value and the characteristics of storage set, has become a very attractive target for attackers. Therefore, big data security is a very important research topic. At present, there are two common methods to guarantee the data security of big data platform (such as Hadoop), which are lack of access control, and there is the risk of being breached by external hackers or bypassed by internal administrators. 2) data encryption. Although high security, encryption and decryption of massive data will increase significant overhead. In order to meet the big data application of high security and high performance requirements for data storage. In this paper, trusted solid state hard disk (SSD) is proposed, which provides secure and enhanced storage device interface and protocol, which enables users to exercise fine-grained access control over the data in storage. In order to ensure the security of data in storage, the security of trusted solid-state hard disk is analyzed in depth, and the challenges and solutions in system design and implementation are introduced in detail. The experimental results show that. On both synthetic and real workloads, trusted solid-state drives can run at less than 3um overhead. Therefore, trusted solid-state drives are expected to become a new basis for big data security.
【作者单位】：清华大学计算机科学与技术系;清华大学信息技术研究院;清华信息科学与技术国家实验室(筹);
【基金】：国家“九七三”重点基础研究发展规划项目基金(2011CB302302) 国家“八六三”高技术研究发展计划项目基金(SS2015AA020102) 清华大学自主科研计划资助
【分类号】：TP333.35;TP309.2
【正文快照】： “八六三”高技术研究发展计划项目基金(SS2015AA020102)和清华大学自主科研计划资助.田洪亮,男,1987年生,博士研究生,主要研究方向为云数据安全、基于新硬件的数据处理.E-mail:tatetian@gmail.com.张勇,男,1973年生,博士,副研究员,主要研究方向为数据管理和数据分析.许信辉,

【相似文献】