Availability Analysis of IP Multimedia Subsystem in Cloud Environments

Abstract

As of today, telecommunication providers are exploiting the possibilities offered by the cloud paradigm to efficiently decouple physical network resources, like hardware equipment, optical interfaces and cables, from offered services, for instance multimedia content delivery and data storage. Among technologies conceived to implement this paradigm, containerization stands out. It can be considered as an evolution of classic virtualization, where software instances called containers are designed to offer specific network functionalities by relying on a separate infrastructure composed of virtual machines and hardware. In line with this new trend, we characterize, from an availability viewpoint, an IP Multimedia Subsystem (IMS) architecture deployed in a containerized environment (dubbed cIMS), which represents a pivotal part of novel network architectures such as 5G. Firstly, we model the availability of cIMS by employing both Reliability Block Diagram (RBD), to capture logical dependencies among cIMS nodes, and Stochastic Reward Networks (SRN), to characterize individually the probabilistic behavior of each node. Then, also supported by an ad-hoc automated procedure, we carry out an experimental assessment of a typical telecommunication network service satisfying a desired availability constraint, whose results are some feasible cIMS configurations that can be deployed.