Realizing Software Reliability in the Face of Infrastructure Instability

Abstract

Cloud computing has brought with it the utilization of off the shelf, commodity hardware that has higher failure rates than the systems that have been used in enterprises for the last several decades. Coupled with increasingly complex, highly-distributed, constantly-changing data center environments that can no longer be treated as deterministic systems, this forces us to change the way that we view the stability of that infrastructure. Our aim is to provide digital solutions that remain stable in the face of this infrastructure uncertainty and we achieve this by utilizing specific design patterns and operational practices. At the core of the new way of working is a philosophical view that change is the rule rather than the exception. US entertainment company Netflix has fully embraced this new mindset and this served them well during a major outage experienced by their cloud infrastructure provider, Amazon Web Services (AWS). In this piece, we study how Netflix was able to avoid any significant impact while many other well established, technically savvy AWS users were not.