Performance prediction of cloud applications through benchmarking and simulation

摘要

As the cloud paradigm gains widespread adoption, the performance evaluation and prediction of cloud applications remain daunting tasks, not yet fully accomplished. Nevertheless, reliable performance figures are the key to take the cloud to the next step, in which it will be possible to predict the maintenance cost of the applications and to introduce richer service level agreements between service providers and consumer. In this paper, we propose a methodology based on benchmarking and simulation that aims at predicting the performance of cloud applications developed through the mOSAIC framework. We prove the efficacy of the methodology on a real case study, showing how it is possible to predict performance indexes (throughput, message queue length, ...) under a generic workload, using pre-acquired benchmark results and simple simulation models.