Research On File Hierarchical Storage Method And Application Of Mass Storage System In Network Library

The computer data is exploding in the context of information. Network repository as a new medium, the stored data is growing amazing. How to organize and save these data effectively and how to control the cost are the main questions the store managers need to study. Actually, storage cost is the first question the enterprise had to resolve. To design a storage system for network repository, how can both reduce the cost of enterprise storage, and as far as possible do not affect the data access efficiencies? This requires a suitable data storage system, and a variety of innovative ways to store date effectively. By doing of these we can reduce the storage complexity, improve the economic benefit.If the managers save all the data in the same storage device, it needs to by more and more storage devices to expansive the storage capacity. If all the data stored in the high-performance storage equipment, the enterprise cannot burden the cost. If they are stored in the low-performance storage device, it will have low business efficiency. In order to take into account the cost and performance, this paper puts forward that use hierarchical storage management theory on network repository storage system.Hierarchical Storage Management theory is based on the Information Lifecycle Management theory. LIM point out that when information produced in a moment it will have a great visits, but over a period of time it is rarely visited, or even visited no. Based on this theory, Hierarchical Storage Management theory indicates that date with different visits should be separate from the different storage devices. The data with great visits that just created should be placed on high-performance storage devices, and the data with small access amounts should be placed on the cheaper and low-performance storage devices. Such a data center needs only a small part of the high-performance storage devices, and for the most part it only need inexpensive storage devices. This not only reduces the cost of storage systems, but also to ensure the performance of the system.Paper first analyzes the theory of hierarchical storage system, and its application status quo and fields, and the methods how computer system organization. Demonstrates it is feasibility to use hierarchical storage on network repository data storage system.Next the paper contrasts the performance and price of the existing mainstream disk, and analysis the main problems of current data center. Then propose hierarchical storage system architecture based on files with three levels: SSD+MAID+automatic type library. Using this system architecture on network repository can reduce the price, ensure the system in the storage system performance, and effectively reduce the energy consumption of the data center.Then according to the properties of the data of network repository, this paper puts forward the dynamic threshold based on the average I/O heat method. This migration algorithm makes an accurate assessment of the file value. At the same time, this algorithm does not need to sort all of the files from the new value every time, so it saves the system resources.