动态最优路径机制在网络存储中的应用

发布时间：2018-03-31 12:55

本文选题：网络存储系统　切入点：Pareto理论　出处：《西安建筑科技大学》2013年硕士论文

【摘要】：随着“大数据”的出现，多元化数据在人类生活中的应用也越来越广泛。这促进了存储技术的不断更新、改变，因此企业及用户要求数据存储应用必须具有高效性以及多元性，所以网络存储系统在“大数据”时代中得到快速发展的空间。系统的高容错性、大容量存储、高网络带宽以及自动化的管理模式将会是网络存储系统的发展目标。在大规模的网络存储系统的设计中，网络资源配置以及存储系统的数据管理策略必然是一个研究重点，因为两者无论在提高用户使用资源满意度还是对于网络存储系统性能的总体提升都会产生重大的影响。在对网络存储系统的研究中，论文通过引入动态最优路径机制，结合网络存储系统的体系结构，构建了基于Pareto理论的网络存储模型；同时，针对网络存储系统的数据分配进行研究，设计了一种基于B+树以及数据分层的数据分配策略，并对研究大规模网络存储系统数据的插入、查询等操作提出了优化的方法。论文的主要内容如下：从用户角度看来，尽管越来越多的网络服务在需求上，基本可以达到用户的要求，但对于网络存储系统这样复杂的环境，就迫切需要一种最优策略帮助用户优化资源配置，提高用户对于网络服务质量QoS，其中通过Pareto最优理论对服务资源的优化配置，提高用户对于网络存储系统的资源偏好，也就是确保用户对于使用网络存储资源的满意度始终保持在满意的状态。虽然Pareto最优理论在处理消费者以及生产者之间的效用均衡问题上有很大的优势，不过对于处理多维数据上还是有不足的地方，所以论文在Pareto理论的基础上通过引入粒子群算法解决出现的多维数据的问题，进一步优化网络存储系统的动态资源配置问题。在网络存储资源配置优化基础上，论文从数据分配策略这方面对网络存储系统进行优化，采用B+树的数据结构方式对网络存储系统中的文件数据进行处理，保证数据读写的稳定性，同时根据数据生命周期的不同，对数据引入了分层搜索的概念，，在满足系统稳定性的前提下，提升了系统对数据读写的速率，优化了数据插入以及查询路径，减少了平均查找时间，以满足用户对网络存储系统的需求。
[Abstract]:With the emergence of "big data", the application of diversified data in human life is becoming more and more extensive. This has promoted the continuous updating and changing of storage technology. Therefore, enterprises and users demand that the application of data storage must be efficient and diversified. So network storage system in the "big data" era of rapid development space, the system of high fault tolerance, large capacity storage, High network bandwidth and automatic management mode will be the development goal of network storage system. In the design of large-scale network storage system, the configuration of network resources and the data management strategy of storage system will be a research focus. Both have a significant impact on the overall performance of the network storage system, both in terms of user satisfaction with the use of the resource and the overall improvement in the performance of the network storage system. In the research of network storage system, by introducing dynamic optimal path mechanism and combining the architecture of network storage system, this paper constructs a network storage model based on Pareto theory. This paper studies the data allocation of network storage system, designs a data allocation strategy based on B-tree and data stratification, and puts forward an optimized method to study the operation of data insertion and query in large-scale network storage system. The main contents of the thesis are as follows:. From the point of view of users, although more and more network services can basically meet the requirements of users, for a complex environment such as network storage system, an optimal strategy is urgently needed to help users optimize resource allocation. In order to improve the QoS of network users, the optimal allocation of service resources is based on the Pareto optimal theory, and the resource preference of users for network storage system is improved. This is to ensure that users' satisfaction with the use of network storage resources is maintained at all times. Although the Pareto optimal theory has great advantages in dealing with the utility equilibrium between consumers and producers, there are still some shortcomings in dealing with multidimensional data. Therefore, based on Pareto theory, particle swarm optimization algorithm is introduced to solve the problem of multi-dimensional data, and the dynamic resource allocation problem of network storage system is further optimized. On the basis of optimizing the configuration of network storage resources, this paper optimizes the network storage system from the aspect of data allocation strategy, and processes the file data in the network storage system by using the data structure of B-tree. At the same time, according to the difference of data life cycle, the concept of hierarchical search is introduced to the data, which improves the rate of data reading and writing under the premise of satisfying the stability of the system. The data insertion and query path are optimized to reduce the average lookup time to meet the needs of users for network storage systems.
【学位授予单位】：西安建筑科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP333

【参考文献】