Availability of distributed systems: A new paradigm

Distributed systems are synonymous with many advanced technologies such as data centers, mobile applications, and the emerging internet of things. These systems require tools for availability assessment. However, traditional models are not equipped to deal with distributed topologies. For example: consider a data center with many servers. What is the critical number of failed servers that defines a system failure? Serial, Parallel, standby, and K out of N models cannot accurately describe the system because server contributions to the data center functionality is additive. In this paper a new approach is presented for availability calculations of distributed systems. The new model accounts for additive contributions to system availability. An example is given as well as comparison between the new approach and traditional models. Furthermore, it will be shown that the new approach is applicable also to many traditional systems such as Oil & Gas facilities, utilities, and plants with multiple production lines. The new approach is easy to implement and can facilitate improved decision making and system designs.