An operating system (OS) is typically composed of a kernel as a resource coordinator that, along with a set of supporting elements (system libraries, user interaction programs etc), interfaces the user space programs with the computing hardware. Since the early UNIX era, the kernel code base has grown significantly. However, the introduction of new code also contain flaws that cause kernel performance degradations, which in turn directly affect the user program throughput, interactivity and efficiency. There is still a paucity of approaches and tools to detect and localize such performance issues in OSs with most existing approaches/tools restricting themselves to performance regressions. The aim of my research is to develop effective strategies for the detection and localization of ernel performance issues to aid realize performant kernels.