The Design And Implementation Of QoS Control System Based On Storage Virtulization System

Quality of Service can provide different service levels for diffenent users, applications or data flow based on its priority. QoS has been widely used in the field of computer network. For the storage system, QoS could also be used to control storage IOs. When the total IOPS demand exceeds the ability that the system can provide, storage QoS can improve the quality of storage service by allocating more resources to high priority IOs. With the rapid development of cloud storage, storage QoS has become more and more important since it can bring many convenient for cloud storage service providers.The research work of this paper is derived from the internship practice project. This project aims at providing more efficient and flexible storage service and controlling IO flow of every virtual disks, by adding QoS control system for a storage virtualization system. In addition, the project also makes high portability requirements for the system. The QoS control system should not depend on a specific hardware or software environment, but can be ported to any storage systems or storage virtualization systems. In view of the above requirements, this paper presents a design scheme of the storage QoS control system based on QoS control technology. The design uses token bucket algorithm to realize the precise control of the IO traffic and minimizes the coupling of storage virtualization system and QoS control system as much as possible so that this system could be easily integrated to other storage system.This paper introduces the architecture design, important data structures of every module and the interfaces of QoS control system in detail. Then make a detail description and explanation of the implementation of the key functions in the system. After the integeration of storage virtualization and the QoS controller system, the design of the virtual disk IO rate control experiment scheme and experimental results are given. The results have proved that the system can control the IOPS and MBPS of every virtual disks accurately. The project achieves the desired requirements of virtual disk 10 traffic control.