(A) design of the direct procedure call IPC scheme for microkernel operating systems

Abstract

Microkernel technology has been the focal point of operating system research for many years and there are quite a few implementation. In microkernel architecture, the internal system service modules of the operating system are separated into multiple independent server programs and run in user space as system server processes. Because of the feature of microkernel architecture, microkernel operating systems need frequent interprocess communications between system server processes to handle users`` service requests. Therefore, most microkernel operating systems suffer from the performance degradation which is called inefficiency problem. Unfortunately, well-known microkernel operating systems have used typical message passing or shared memory IPC scheme as the IPC facility. In this paper we propose the Direct Procedure Call(DPC) IPC scheme for interactions among the operating systems components, which are the microkernel and the system servers, to improve system performance. We also describe necessary system design issues to implement this scheme. The DPC IPC scheme does not have any processing elements on IPC flow. Also for the other elements, overhead is decreased. Conclusively, because the required time for DPC IPC is the same as that of local procedure call at the best case or is shorter than that of other IPC schemes, we confirmed that the DPC IPC scheme is one solution that can improve the performance of the microkernel systems.