metachory

Virtualization has provided a vehicle for people to gain flexibility and security in utilizing computing resources. At the same time, distributed systems have emerged to support large workloads and efficient use of clustered resources. Such distributed systems are often highly specialized and require considerable effort on behalf of the application developer, end-user, system administrator or some combination of all three to use. As a result, it is difficult, if not impossible, for ordinary end-users to take advantage of large-scale distributed computing.

To address these problems, we have created a new OS kernel that virtualizes only the exceptional aspects of native program execution such as system calls, page faults, signals, and file system interactions. Combined with a distributed coherence mechanism, our kernel proves to be a feasible approach to creating a general purpose distributed system that is not only simple to install, configure, and maintain but, more importantly, easy to use. Our approach provides a foundation for supporting execution of unmodified native applications - at nearly-native speed - by ordinary end-users without administrative or special privileges (e.g., "root") as well as transparent process migration and checkpointing.

Metachory was originally conceived in 2004 by Rick Kennell. Work proceeded for about a year before pausing due to lack of time. The project restarted in 2008 with Jeff Turkstra and began a period of intensive development. Metachory represents the culmination of Jeff's doctoral thesis and was made available in May of 2013 as part of his dissertation.

The name metachory comes from a latinization of the made-up Greek word "μεταχωρη." This word is composed of the prefix "meta-" meaning "beyond", "after", or "transcend" and the root "chory" (or "choric") meaning "space" or "place." Combining the two results in "metachory" meaning roughly "transcendence of place."