This is a talk about changes in the nature of computing that have the potential of dramatically affecting the architecture of operating systems, which have been relatively stagnant for decades. A general breakdown in kernel abstractions, the coming plethora of CPUs/socket (i.e., “ManyCore”), and virtualization, all represent inflection points that may cause a fundamental re-think. None of the discussion is about specific product plans, but is instead a high-level look at what operating systems may look like in the future, and why we may be about to enter one of the most revolutionary eras in the design and implementation of the traditional “kernel” layer of the OS. It is possible that kernels, as we know them today, may become obsolete.
Dave Probert is an architect in the Windows Core Kernel & Architecture team at Microsoft. Until recently Dave managed kernel development for Windows 2000, XP, Server 2003, and early phases of XP SP2 and Vista. Currently Dave is working on releasing kernel sources to universities and developing Project OZ, an experimental environment based on the SPACE project at UC Santa Barbara, where Dave earned his Ph.D. in Electrical & Computer Engineering.
Prior to Microsoft, Dave's experience was primarily focused on Unix kernels, including several years as Vice President of Software Engineering at Culler Scientific Systems. Dave's career began in the late 1970s at Burroughs, where he was a computer architect designing hardware and writing microcode for the B1900.
Wu-chi Feng