Method Study On Improving Performance Of Massive Magnetic Recording System

Currently, information industry based on computer and network has got an unprecedented development. The world has entered the information age, the outstanding characterists of which are as follows: the explosive growth of the amount of information and the amount of information exchange, the emerging of many new application areas (such as video on demand, scientific data processing, data centers, etc.), causing the demand for the data storage to grow at a several-fold rate every year, which rises a increasingly higher demand on information storage technology.In all the ways of information storage, the magnetic recording, due to its excellent recording performance, flexible application, lower cost and its great potential in the technology development, is regarded as a major modern information storage technology. Magnetic recording storage of massive data is the most commonly used massive storage mode.This paper chooses "Method Study on Improving Performance of Massive Magnetic Recording System," which has not only scientific values but also practical significance, with the purspose of providing a theoretical basis and optimization methods to improve the performance of massive magnetic recording systems. Currently, the use of magnetic disk arrays to form the massive magnetic recording system has become an inevitable trend, and disk array-related technology has become a hot area of research and the mainstream technology for data storage. Disk arrays has become an important storage system architecture.The main contents of this paper are the following:1. Massive magnetic recording system construction technology. Disk array technology uses different combinations of multiple hard drives to form a large capacity, fast responding, high reliability storage system, which can achieve the parallel operation of multiple disk drives, greatly improving the storage capacity and data transfer rate, improving its reliability by the redundant backup technology. Therefore, this paper uses disk arrays technology in building this massive magnetic recording system, studies the basic principles for the disk array, disk array RAID level classification and its characteristics together with the common data channel of storage servers, and finally studies implementation technology of software RAID and hardware RAID.2. Massive magnetic recording system modeling technique. It mainly foucs on how to reduce seek distance and disk access delay and improve throughput by disk array configuration. This paper discusses the basic principles for disk array modeling and a variety of techniques to improve the system latency using disk arrays, suggests a new method of designing disk arrays, which can flexibly and systemetically reduce the disk access latency, increase disk throughput, and the new disk array is called as "SR array", because it can evenly minimize seek time and rotation delay, and this paper, based on it, discusses the details of the disk array configuration, deduce the SR delay model and its throughput model.3. Disk head tracking software design. Disk head tracking software does not require any specialized hardware support in high-precision positioning and reducing overhead, while it achieves a variety of scheduling algorithms with a heavy load, such as the shortest access time first (SATF) algorithm. This paper gives the head tracking software layered architecture and its software design methods, uses different scheduling strategies to implement a software test platform that supports diverse SR array configuration, and integrates a simulator in the experimental platform, simulation results show that simulator can truly reflect the actual operation of the system read and write requests for implementation.4. Massive magnetic recording system performance test simulation. Use disk performance testing tool software IOMeter to test the emulator integrated in the system. Event queue within the emulator is based on chronological order of the events. Emulator can record the event processing time, and can adjust the clock according to the processing reality. The emulator can simulate the actual system seek and rotation delay. Simulation results show that among all possible SR array configurations, the difference between the simulator and the prototype system is less than 3%. This paper attempts to verify the accuracy of the SR delay array model and throughput model, compare the simulation results of the emulator under different workloads. The results show that the mathematical model result is indeed very close to the actual measurements.