Practical Aspects of IP Take-Over Mechanisms

Abstract

Transparent replication has been viewed as the holy grail of fault-tolerant computing. We discuss issues arising when using atomic broadcast to replicate services that are accessed via the Internet. We show how some of these issues can be addressed by the use of IP based replication schemes. In particular, we argue for the replication of services using a simple IP take-over scheme. Also, we show how a simple fail-over scheme can be combined with a load shedding and balancing mechanism.