Availability assessment and sensitivity analysis of an MBaaS platform
Abstract
The OpenMobster platform offers services for the mobile cloud in a complete way. However, OpenMobster's availability requires attention. Analytical models are usually used to evaluate the dependability of a system, enabling the reduction of downtime and other advantages. This work proposes a set of stochastic Petri net models focused on evaluating the availability and reliability of the MBaaS OpenMobster platform. A sensitivity analysis was performed to identify the system's most critical components. The base model obtained an availability considered low, 96.8%. An extended cold-standby model with server redundancy was implemented, resulting in a better availability, 97.09%. Due to the low significant increase in availability, another redundancy strategy was applied to the MBaaS service model. A self-healing technique was used, which presented the best availability among the three proposed models with 99.91%.