Benchmarking autonomic capabilities: Promises and pitfalls

Abstract

The advantages and disadvantages of benchmarking automatic computing are discussed. It is stated that developing appropriate benchmark for autonomic computing helps to quantify the progress towards self-managing systems. An autonomic benchmark must supplement the performance workload with a representative, reproducible set of changes to inject. One of the disadvantages of the system is the requirement of large-scale simulation infrastructure, for effecting self-protection, self-healing and self-configuration, which makes the real deployment very complex.