Research and optimization on quality of service in distributed block storage system

In cloud storage technology,resource reuse and sharing is a double-edged sword.While it reducing the overhead for customers,it leads to a situation that the different applications will compete for hardware resources,especially when resource requirement exceeded storage system capability.As a consequence the cloud storage system is unable to provide satisfactory quality of service.An adaptive scheduling technology was proposed,solving the resource balance problem between front tasks and back tasks.A multilevel popularity cache replacement technology was also proposed,solving the resource balance problems between the front tasks themselves.These technology was implemented and tested on the ZTE distributed block storage system.The proposed technology satisfies the expectations of the quality of service from the users.