Hierarchical Storage Structure And Algorithms

As the global continuous development of information technology and digitization, the computer system applications in various industries continue to grow with the application types increasing. At the same time, with the increase of the number application types, the number of users is increasing, so the amount of data generated by users is also growing. All these reasons lead to explosive growth in the amount of data. Large amounts of data need to store and access, which has made the computer storage system under heavy pressure. In order to store the rapid growing data, storage technology had a rapid development in recent years. The performance of equipments was continuously improved, resulting in a number of new storage structures and techniques. To a certain extent, the technology development eased the pressure on data storage, but it has brought some other problems like the over high costs of storage, inconvenience data management and so on. In this context, IBM proposed the idea:hierarchical storage management(HSM), which can make storage devices play performance effectively, improve data access efficiency and reduce storage costs. This concept which has broad market prospects has been widely recognized and sought after manufacturers, as current research and development of hot spot.First, the thesis introduced the historical development of hierarchical storage ideas and the origin of concepts. Then it introduced the hierarchical storage current research overview and some problems in current practical application. Through the analysis of several basic storage architectures, the thesis compared these different characteristics of HSA which were built upon the basic architectures. These works confirmed SAN architecture had many advantages. The thesis analyzed the different requirements for storage in several practical typical application modes. On this basis, it described a comprehensive study of applications: university library HSA design and implementation. In the process of hierarchical storage architecture design, two methods were proposed. One was a storage level classification method which was combined with application, the other was a storage space determination method. Both of them were used in the design process of the HSA project. According to practical verification, these two methods did not affect the data access. Both of two were effective solution to the problem that the traditional three-level division of hierarchical storage did not suit to the comprehensive applications. Using a quantitative method which enhanced the effective utilization of all hierarchical storage devices could further reduce the storage costs.Then the thesis analyzed two typical kinds of data classification algorithm. Based on the analysis, the algorithm was improved by combining advantages of both the typical algorithms. It was named the dynamic threshold method of data classification. What is next was briefly implementation of the new algorithm. According to the characteristics of university library data access and Information Lifecycle Theory, a preliminary study of the data migration strategy was done. At the end, two kinds of data migration strategies which were access frequency-based were proposed:the periodicity strategy and the predictability strategy. These strategies effectively improved the traditional which maybe made data distribution unreasonable or a lag of data migration in some circumstances of data access.