基于IP-SAN的智能固态缓存网络存储系统的研究

发布时间：2018-01-10 12:12

本文关键词：基于IP-SAN的智能固态缓存网络存储系统的研究　出处：《南京林业大学》2012年硕士论文　论文类型：学位论文

【摘要】：随着云存储时代的到来，，网络数据存储不仅仅在容量上呈爆炸式增长，在性能和可靠性上也有越来越高的要求。而现今的机械硬盘构建的SAN存储系统，在大容量以及顺序读写方面虽然具有很大优势，但已经无法满足社会发展所带来的新的需求，例如虚拟技术以及基于数据库的视频点播、电子商务等应用，对快速随机访问要求很高。固态硬盘以其高随机性能、超低延时等优势倍受青睐，但其高昂的价格让存储界望尘莫及。鉴于此，本文研究并实现了智能固态缓存技术，提出了基于IP-SAN的具有高随机性能的智能固态缓存网络存储系统解决方案。本课题主要研究内容如下： 1、阐述了现阶段国内外对随机性能的研究状况，研究了主流网络存储架构特点。研究了RAID技术，分析和比较了不同RAID级别实现机制及其特点。 2、详细研究了linux内核的块设备和设备映射机制，分析了linux内核中数据的I/O请求方式和实现机制，研究了linux内核实现逻辑设备与物理设备的映射过程。详细介绍了映射设备、映射目标和映射表三者的结构和它们之间的关系。基于此，设计并实现了智能固态缓存技术，极大的提升了存储系统的随机性能。着重阐述了智能固体缓存的逻辑架构和实现机制，介绍了其创建方法和各功能函数的具体实现，并在本地磁盘阵列上运用该智能固态缓存功能对随机读写性能进行了比较。 3、研究了网络存储系统中的逻辑卷管理和iSCSI协议的实现。举例说明了逻辑卷管理的实现过程，研究了iSCSI的体系结构、错误恢复机制，以及会话实现过程。iSCSI网络采用双网卡聚合，研究了网卡聚合的原理和创建方法，以实现负载均衡和故障切换技术，进一步提升存储系统的性能。 4、最后对整个网络存储系统的功能、性能和稳定性进行了全面详细的测试。并对测试结果进行了比较和分析。
[Abstract]:With the arrival of cloud storage era, network data storage not only has explosive growth in capacity, but also has more and more high requirements in performance and reliability. Nowadays, SAN storage system is constructed by mechanical hard disk. Although it has great advantages in large capacity and sequential reading and writing, it can not meet the new needs brought by social development, such as virtual technology, database based video on demand, e-commerce and other applications. The demand for fast random access is very high. Solid state hard disk is very popular for its advantages of high random performance and ultra-low delay, but its high price makes the storage world out of reach. In view of this, this paper studies and implements the intelligent solid state cache technology. An intelligent solid state cache network storage system with high random performance based on IP-SAN is proposed. The main contents of this paper are as follows: 1. The research status of random performance at home and abroad, the characteristics of mainstream network storage architecture, and the RAID technology are described. The implementation mechanism and characteristics of different RAID levels are analyzed and compared. 2. The block device and device mapping mechanism of linux kernel is studied in detail, and the I / O request mode and implementation mechanism of data in linux kernel are analyzed. The mapping process between logical devices and physical devices in linux kernel is studied. The structure of mapping devices, mapping targets and mapping tables and their relationships are introduced in detail. The intelligent solid state cache technology is designed and implemented, which greatly improves the random performance of the storage system. The logical architecture and implementation mechanism of the intelligent solid state cache are described in detail. The creation method and the realization of each function function are introduced, and the performance of random reading and writing is compared by using the intelligent solid state cache function on the local disk array. 3. The realization of logical volume management and iSCSI protocol in network storage system is studied, the implementation process of logical volume management is illustrated, and the architecture of iSCSI and the mechanism of error recovery are studied. The principle and creation method of network card aggregation are studied in order to realize load balancing and failover technology and further improve the performance of storage system. Finally, the function, performance and stability of the whole network storage system are tested in detail, and the test results are compared and analyzed.
【学位授予单位】：南京林业大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP333

【参考文献】