A CASE FOR 2-LEVEL RECOVERY SCHEMES

Long-running applications are often subject to failures. Failures can result in significant loss of computation. Therefore, it is necessary to use a failure recovery scheme to minimize performance overhead in t he presence of failures. In this paper, we argue that it is often adva ntageous to use ''two-level'' recovery schemes. A two-level recovery s cheme tolerates the more probable failures with low performance overhe ad, while the less probable failures may possibly incur a higher overh ead. By minimizing overhead for the more frequently occurring failure scenarios, the two-level approach can achieve lower performance overhe ad (on average) as compared to existing recovery schemes. The paper de scribes two two-level recovery schemes. Performance analysis using a M arkov chain shows that, in practice, a two-level scheme can perform be tter than its ''one-level'' counterpart. While the conclusions of this paper are intuitive, the work on design of appropriate recovery schem es is lacking. The objective of this paper is to motivate research int o recovery schemes that can provide multiple levels of fault tolerance and achieve better performance than existing recovery schemes. The pa per presents an analytical approach for evaluating performance of two- level schemes and shows that such schemes are hard to optimize analyti cally.