Towards Recoverability-Oriented Cloud Operations

The dependability of consumer-initiated cloud operations has become a hot research topic in the field of cloud computing. Such a cloud operation's dependability can be achieved by enabling its recoverability, e.g. applying recovery actions for errors during cloud operations. In order to facilitate cloud operations' recoverability, one way is to design more recoverability-oriented cloud operations. Cloud operations designed with more effective recoverability-oriented characteristics can receive recovery actions by a recovery service with ease. In this paper, we propose a methodology of how to make such kind of cloud operations. Specifically, the methodology contains three steps: 1) operation prototype design; 2) cloud resources analysis; 3) operation prototype transformation. We demonstrate our methodology by using the widely used rolling upgrade operation on AWS Cloud.