> Hypervisors just seem like particularly impoverished microkernels to me. Or, "Microkernels done right": https://www.usenix.org/legacy/event/hotos05/final_pa

> Hypervisors just seem like particularly impoverished microkernels to me. Or, "Microkernels done right": https://www.usenix.org/legacy/event/hotos05/final_papers/full_papers/hand/hand.pdf https://www.usenix.org/legacy/event/hotos05/final_papers/ful... > At the last HotOS, Mendel Rosenblum gave an ‘outrageous’ opinion that the academic obsession with microkernels during the past two decades produced many publications but little impact. He argued that virtual machine monitors (VMMs) had had considerably more practical uptake, despite—or perhaps due to—being principally developed by industry. > In this paper, we investigate this claim in light of our experiences in developing the Xen [1] virtual machine monitor. We argue that modern VMMs present a practical platform which allows the development and deployment of innovative systems research: in essence, VMMs are microkernels done right. (Edited to add relevant paragraphs from the abstract.)