VoD in eucalyptus platform: Availability modeling and sensibility analysis

Due to advantages such as the pay-per-use model and a highly scalable environment, cloud infrastructures are commonly employed by Video on Demand (VoD) streaming services to supply an extensive range of multimedia content to users. However, recent events have demonstrated to commercial VoD service providers the fundamental importance of ensuring highly available cloud computing infrastructures. This paper examines the effectiveness of applying availability modeling to a VoD cloud environment which has no redundancy mechanisms included in its architecture. The analysis employs a combinatorial modeling strategy consisting of RDB models and Markov chains. Sensitivity analysis is also applied to identify the critical points in the system with respect to availability. The outcome of the research clearly shows that the modeling strategy, together with the sensitivity analysis, has a key role to play in the identification of those components on which resources should be focused in order to achieve increased system availability.