A survey on auto-scaling: how to exploit cloud elasticity

Abstract

Elasticity plays an essential role as far as the wide diffusion of cloud computing is concerned. It enables a cloud application deployment to 'scale' automatically, adapting to workload changes, guaranteeing the performance requirements with minimum infrastructure leasing costs. However, auto-scaling poses challenging problems. This paper gives a detailed overview of the current state of the art on auto-scaling. Firstly, the key design points for auto-scaling tools are presented and discussed. Then, literature proposals and on-going research are dealt with. Finally, existing auto-scaling implementations, including those used by commercial cloud providers, are reviewed.