分级存储系统中数据迁移技术的研究

发布时间：2019-03-18 10:50

【摘要】：随着信息化、数字化的不断发展，产生数据的速率呈指数增长。大量数据需要存储和访问，给存储系统带来了极大的负担。为了存储急速增长的数据，近年来存储技术在快速发展，存储设备的性能也在不断提高，涌现出了许多新的存储结构和技术。这些技术虽然在一定程度上缓解了存储的压力，但与此同时也带来了存储成本的大幅度提高，使存储管理更加复杂。在这种情况下，分级存储作为存储领域的一个重要角色受到了越来越多的关注。分级存储思想作为一种能够有效发挥存储设备性能，提高数据访问率以及降低存储成本的概念，有着非常广阔的应用前景，成为当前研究的热点。分级存储的核心是数据迁移技术，该技术在不同存储层次之间迁移数据，同时保证迁移过程中数据访问的一致性。本文对分级存储系统结构中的数据迁移技术展开研究。具体的研究内容如下：首先，研究了当前主流的网络存储存储技术；详细对比分析了磁盘阵列、固态盘、虚拟磁带库、光盘等存储设备当前的性能，为后边分级存储平台的构建作一个铺垫；介绍了分级存储的的概念、必要性、优点、方式和应遵循的规则。其次，研究了当前的数据迁移策略，基于存储空间的高低水位法和基于数据访问率的Cache替换迁移算法，分析了这些数据迁移策略的不足之处；基于信息生命周期管理的思想，提出了MSDV（MigrationStrategyBasedonDataValue）数据价值评估模型，该模型与数据迁移过程控制策略相结合可以在尽量减小对系统影响的情况下进一步提高访问高性能设备的命中率。最后，在Disksim模拟器上，对上述迁移算法进行了性能测试，将基于MSDV的数据迁移控制策略与Cache替换迁移算法（LRU、LFU、LRU-K和LFU-Aging）进行了对比研究，，验证了此迁移策略的高效性。
[Abstract]:With the development of informatization and digitalization, the rate of producing data increases exponentially. A large amount of data needs to be stored and accessed, which brings a great burden to the storage system. In order to store the fast-growing data, the storage technology is developing rapidly in recent years, and the performance of the storage devices is improving constantly. Many new storage structures and technologies have emerged. Although these technologies alleviate the pressure of storage to a certain extent, but at the same time, they also bring about a large increase in storage cost, which makes the storage management more complex. In this case, hierarchical storage as an important role in the field of storage has received more and more attention. Hierarchical storage, as a concept that can effectively exert the performance of storage devices, improve the data access rate and reduce the storage cost, has a very broad application prospect, and has become a hot research topic at present. The core of hierarchical storage is data migration, which migrates data between different storage levels and ensures consistency of data access during migration. In this paper, the data migration technology in hierarchical storage system architecture is studied. The specific research contents are as follows: firstly, the current mainstream network storage technology is studied; The current performance of storage devices such as disk array, solid state disk, virtual tape library, optical disk and so on is compared and analyzed in detail, which lays the groundwork for the construction of the rear hierarchical storage platform. This paper introduces the concept, necessity, advantages, ways and rules of hierarchical storage. Secondly, the current data migration strategies, high and low water level method based on storage space and Cache replacement migration algorithm based on data access rate are studied, and the shortcomings of these data migration strategies are analyzed. Based on the idea of information lifecycle management, a MSDV (MigrationStrategyBasedonDataValue) data value assessment model is proposed. The combination of the model and the data migration process control strategy can further improve the hit ratio of accessing high-performance devices with minimal impact on the system. Finally, the performance of the migration algorithm is tested on the Disksim simulator. The data migration control strategy based on MSDV is compared with the Cache replacement migration algorithm (LRU,LFU,LRU-K and LFU-Aging). The efficiency of this migration strategy is verified.
【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】：2012
【分类号】：TP333

【参考文献】