Uninterrupted uptime is a critical aspect of Virtual Machines (VMs) offered by cloud hosting providers. Google's VMs run on top of rapidly changing infrastructure: we regularly update hardware and host software, and we must quickly respond to failing hardware. Frequent change is critical to both development velocity---deploying new versions of services and infrastructure---and the ability to respond rapidly to defects, including critical security fixes. Typically these updates would be disruptive, resulting in VM termination or restart. In this paper we present how we use VM live migration at scale to eliminate this disruption with minimal impact to the guest, performing over 1,000,0001migrations monthly in our production fleet, with 50ms median blackout, 300ms 99th percentile blackout.

Object-oriented models are a popular basis for supporting uniform sharing of data and services in operating systems, distributed programming systems, and database systems. We term systems that use objects for these purposes object sharing systems. Operating systems in common use have nonuniform addressing models, making the uniform object naming required by object sharing systems expensive and difficult to implement. We argue that emerging 64-bit architectures make it practical to support uniform naming at the virtual addressing level, eliminating a key implementation problem for object sharing systems. We describe facilities for object-based sharing of persistent data and services in Opal, an operating system we are developing for paged 64-bit architectures. The distinctive feature of Opal is that object This paper will appear in identical form in the proceedings of the Conference on Object-Oriented Programming Systems, Languages, and Applications (OOPSLA), October 1992. This work w...