Who Watches the Watchmen? Protecting Operating System Reliability Mechanisms

We present the design and initial evaluation of a resilient operating system architecture that leverages HW architectures combining few resilient with many non-resilient CPU cores. To this end, we build our system around a Reliable Computing Base (RCB) consisting of those software components that must work for reliable operation, and run the RCB on the resilient cores. The remainder of the system runs replicated on unreliable cores. Our system’s RCB consists of an L4 microkernel, a runtime environment and a replication manager. In this paper we state and justify assumptions about the hardware architecture, motivate the corresponding software architecture and evaluate communication mechanisms between the RCB and the replicas.