Thesis Proposal: Reducing Operating System Intrusion in the Piglet Os
Preliminary results obtained from the current implementation of Piglet show that applications can indeed access system functions with much lower overheads, providing proof of the feasibility of the Piglet architecture and further research needs to be conducted to more fully evaluate it.
Abstract
Operating system intrusion|the overhead imposed upon applications by the operating system itself|has been studied extensively in the OS research community. Consideration of the forms and origins of intrusion reveals that a high degree of intrusion is inherent in a conventional OS due to its structure and passive nature. Thus, new OS architectures are required if the degree of intrusion is to be signiicantly reduced. Piglet, a new multiprocessor operating system architecture, successfully reduces intrusion by making the OS kernel active rather than passive. This permits much less intrusive protection mechanisms to be employed in the construction of the OS, leading in turn to greatly reduced overheads for invo-cation of system services and hence increased application performance. Preliminary results obtained from the current implementation of Piglet show that applications can indeed access system functions with much lower overheads. While thus providing proof of the feasibility of the Piglet architecture, further research needs to be conducted to more fully evaluate it. Among the topics to be investigated are: the implications of asynchronous vs. synchronous kernel structure eecient implementation of user-level services e.g., network protocols which application domains Piglet is most applicable to scalability of the Piglet architecture