Phoenix Application Recovery Project

High availability for both data and applications is rapidly becoming a business requirement. Yet even after decades of software engineering research, computer systems and production applications still fail [11], primarily due to nondeterministic bugs that can typically be resolved by simply rebooting the system or restarting the application [6]. The Phoenix project takes the position that such hardware faults and software failures are facts to be coped with, not problems to be solved. This position is supported both by historical evidence and by recent studies on the primary sources of outages in production systems [11]. Conceding that Heisenbugs will remain a fact of life in computing, the goal of the Phoenix project is to enable applications to survive system crashes, without requiring application developers to take special measures for state persistence and application recovery. This simplifies application programming, reduces operational costs, masks failures from users, and increases application availability. In this paper we describe the principles behind the Phoenix project and outline some research areas and recent projects.