云计算环境下安全分布式存储架构与容错技术研究

发布时间：2018-01-03 04:04

本文关键词：云计算环境下安全分布式存储架构与容错技术研究　出处：《解放军信息工程大学》2013年博士论文　论文类型：学位论文

【摘要】：云计算已经得到广泛的关注,并且发展迅速。以数据中心网络为基础的分布式存储是构建云计算的物理实体。但是由于云计算环境下分布式存储的开放性带来的安全隐患,以及其数据可靠性的制约,使如何实现在数据可容错的过程中保证数据的安全性成为亟待解决的问题。本文研究了云计算环境下安全分布式存储架构与容错技术,主要内容和贡献如下。 1.基于分层源地址验证技术的数据中心网络安全架构本文提出一种基于分层源地址验证的数据中心网络安全架构,设计了一种可验证源地址生成方法。数据中心网络中的服务器在发送数据时必须使用这种可验证地址作为源地址,在数据发送过程中,该地址将会被验证,确保每台服务器都无法仿冒其他服务器进行数据传输,而且从互联网通过开放端口传入数据中心内部的数据在没有被分配可验证地址的情况下无法在数据中心内部进行转发。为了保证验证的高效性,在验证过程中引入了分层验证和流认证的思想,在数据中心内部使用分层验证,而在数据中心间使用流认证。该架构的使用,可以帮助系统发现不正常的数据传输,过滤非法主机在网络内的数据传输,并定位数据中心内部可能的攻击者,防范直接利用互联网上的主机攻击系统获取数据。实验结果表明基于源地址验证的数据中心网络安全架构可以在不影响数据传输的情况下实现对数据包源地址的验证,其已经具备了实用价值,能够解决网络中利用伪造地址对数据网络中心进行攻击的问题。 2.基于广播加密思想的安全再生码本文提出了一种将广播加密模型与再生码模型相结合的安全再生码——FCBE (Fault-tolerant Code Based on Broadcast Encryption,FCBE)。在FCBE模型构建过程中,借鉴了广播加密的思想,将编码存储及数据恢复的过程归结为一个广播过程,在数据存入系统时,由系统为其选择一个安全服务器的集合作为容错服务器。当存储服务器失效时,只有系统选定的容错服务器才能够实现数据恢复,而其他服务器即使截获了发送给容错服务器的数据块,也无法恢复原始数据。安全性分析证明了FCBE能够实现适应性安全,实验结果表明,其引入安全要素所造成的带宽占用是可以接受的,不会对整个数据中心网络的数据传输造成压力。 3.基于门限机制的安全再生码本文提出了两种基于门限机制的安全再生码,其核心思想是在再生码模型中引入可靠第三方密钥服务器,用户将数据存放于数据中心时从编码矩阵中选取部分秘密,并将秘密分享给第三方密钥服务器,当需要对失效节点中的数据进行恢复或者是其他数据使用者下载数据时,需要经过第三方密钥服务器的验证,通过验证后才能够从中获取编码矩阵的秘密,进而构造解码矩阵恢复失效数据或下载原始数据。基于此思想本文提出了两种安全再生码SRCF (Secure Regenerating Code for Fault-tolerant, SRCF)和SRCS(Secure Regenerating code with Semi-adaptive, SRCS)。安全性分析证明了SRCF可以实现选择明文安全,而SRCS可以实现部分适应性攻击安全；实验结果表明,SRCF和SRCS引入安全要素所造成的带宽占用不大,不会对整个数据中心网络的数据传输造成压力。 4.基于流水线思想的编码数据恢复模式本文提出了一种基于流水线思想的数据恢复模式。该模式借鉴了工业生产中流水线生产的思想,将待恢复服务器看作流水线上的产品,将存储服务器看作流水线工人,完成数据恢复的服务器即为生产完毕的产品。利用该模式可以进一步的降低数据恢复时所占用的带宽,从而减小引入安全要素所增加的带宽消耗。通过理论分析证明了该模式不会对数据恢复的正确性造成影响,且其可以减小带宽消耗。
[Abstract]:Cloud computing has received widespread attention and rapid development. In the distributed storage of data center network based cloud computing is to build physical entities. But because cloud computing brings security risks of open distributed storage environment, and restrict the reliability of the data, so how to achieve in the process of data fault tolerance in safety the data has become an urgent problem. This paper studies the cloud computing environment security distributed storage architecture and fault tolerance technology, the main contents and contributions are as follows.
1. data center network security architecture based on layered source address verification technology
This paper presents a data center network security architecture layered source address validation based on the design of a verifiable source address generation method. Data center network server must use this address as the source address validation in sending data, in the data transmission process, the address will be verified, to ensure that each server can not fake other servers for data transmission, but also from the Internet through the open port incoming data in data centers has not been assigned can verify address under the condition of not in the data center for forwarding. In order to ensure efficient verification, the introduction of ideological hierarchical verification in the verification process and stream authentication, in the data center for internal use layered verification, using stream authentication in the data center. The use of the framework, can help the system to find the data transmission is not normal, filtering illegal The host within the network data transmission, and the internal data center positioning possible attacker, attack the host systems on the Internet directly using the prevention data acquisition. The experimental results of data center network security architecture of BenQ in the source address validation can be implemented to verify the package source address of the data did not affect the data transmission, it already has the practical value, can solve the attack on the network data center using forged address problems in the network.
2. secure regenerated code based on broadcast encryption
This paper presents a combination of broadcast encryption model and security model code regeneration (Fault-tolerant Code regeneration code - FCBE Based on Broadcast Encryption, FCBE FCBE). In the process of building the model, from the broadcast encryption theory, encoding storage and data recovery process down to a broadcast in the process, the data is stored in the when the system is set by the system for the selection of a security server as a fault-tolerant server. When the storage server fails, only the selected system fault-tolerant server can realize data recovery, while the other server even if intercepted sent to the data block fault-tolerant server, can not restore the original data. The security analysis proves that FCBE can achieve adaptive security, the experimental results show that the introduction of safety factors caused by the bandwidth is acceptable, the number of the entire data center network will not Pressure is caused by transmission.
3. secure regeneration code based on threshold mechanism
This paper presents two kinds of regeneration code security based on threshold mechanism, its core idea is to model the introduction of regenerating codes for a reliable third party key server, user data stored in the data center is selected from the secret encoding matrix, and the secret sharing to the third party key server, when the need for node failure the restoration of data or other data users to download data, need to go through third party verification key server, through the verification to be able to get from the secret encoding matrix, then structure the decoding matrix failure recovery data or download the original data. The idea of this paper is based on two security code regeneration SRCF is proposed (Secure Regenerating Code for Fault-tolerant SRCS (Secure, SRCF) and Regenerating code with Semi-adaptive, SRCS). The security analysis proves that SRCF can realize the selection of Ming Wenan, SRC S can achieve partially adaptive attack security. The experimental results show that the bandwidth consumption caused by the introduction of security elements by SRCF and SRCS does not occupy much pressure on data transmission in the entire data center network.
4. coded data recovery mode based on Pipelining
This paper presents a recovery model based on the data of the ideological line. From the mode of industrial production in the production line of thought, to restore the server as the product line, the storage server as assembly line workers, complete data recovery server is the production finished products can be further reduced when occupied by data recovery the bandwidth utilization mode, thereby reducing the consumption of security elements increased bandwidth. Through theoretical analysis proves that the model will not affect the correctness of data recovery, and it can reduce the bandwidth consumption.

【学位授予单位】：解放军信息工程大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：TP333;TP393.08

【参考文献】