BECloud: A new approach to analyse elasticity enablers of cloud services

• Elasticity is a key performance figure in cloud computing.
• But there is not a standard metric or procedure to quantify it and it is rarely used.
• We define a new elasticity metric, general, flexible, simple and easy to measure.
• An analysis procedure and a benchmarking tool, BECloud, are also proposed.
• These definitions allow providers and users to analyse service elasticity enablers.

Elasticity is a key property of cloud computing but there is a lack of standard elasticity metrics or analysis procedures to easily quantify this performance figure of cloud services. This absence of a unique general elasticity metric makes difficult to consider elasticity as a service level objective in Service Level Agreements, to benchmark cloud services or to explicitly improve the elasticity of scaling and provisioning mechanisms, to mention only some examples. This paper defines a new elasticity metric capable of considering its four main components, scalability, accuracy, time and cost, independently of the service level (infrastructure, platform or software). Furthermore, an analysis procedure to evaluate the behaviour of service elasticity and a benchmarking tool to automate this analysis are presented. The main elasticity enablers of cloud services are identified and analysed using this metric, procedure and tool via real use cases on private and public clouds, drawing interesting conclusions about this important performance aspect of cloud services.