Aqua: An Adaptive End-System Quality of Service Architecture

Abstract

In this chapter we propose an integrated resource management architecture for an end-system that can deliver desired QOS to multimedia applications. The proposed architecture (called AQUA for Adaptive End-system Quality of Service Architecture) includes a common framework for managing resources such as CPU, network interface, memory, and bus bandwidth. Significant and novel contributions of AQUA include an adaptation framework, QOS specification, resource managers, and an application-level QOS manager that performs application-based graceful adaptation when resource requirements change or the demand for resources exceeds available capacity. We have applied AQUA to the problem of CPU mangement for multimedia applications. The chapter describes a CPU management algorithm called RAP (Rate-based Adjustable Priority Scheduling) that provides predictable service and dynamic QOS control. RAP uses a new admission control test for processes with unknown compute times and methods to use available slack in order to handle processes with varying compute times. Finally, we present results based on a trace-driven simulator that demonstrate the effectiveness of algorithms described in this chapter.