Protection and abstraction are traditional advantages of operating system that provide for application-pro-grammer. How to change Protection and abstraction in order to ensure the new Extensible Operating Systems havinghigh performance and good flexibility is the most crucial problem to build Extensible Operating Systems. We bringforward a new approach--Protected Decentralizing abstractions, integrating Decentralizing OS abstractions and Pro-tected abstractions, and use it to implement a prototype library operating system -- EXLinux/LibOS . A novel wayto safely share user-level abstractions in the prototype, enables unprivileged and untrusted applications to be definedand securely shared generic abstractions at run-time ,in which the same abstraction is invoked repeatedly.
Operating systems research has traditionally consisted of adding functions to the operating system or inventing and study new implementations. Regardless of the research goal, the single constant has been that the size and complexity of operating system increase over time. As a result, operating systems become the most single complex piece of software in a computer system. Today's operating system research is aimed at finding new ways to construct the operating system in order to increase its flexibility, allowing it to adapt to change in designing such extensible systems and the array of choices facing the operating system designer. We discuss such question for describing extensible operating systems and then advance four key design point for extensible operating systems, we analyze and argue each point that relate to different extensible technology. Provide scientific methodology for Operaring systems research or development.
