Big data analytics on container-orchestrated systems

Abstract

Container-orchestration systems offer new possibilites to software architects seeking to make their software systems more scalable and reliable. In the past, these systems have been used to implement transactional software systems but, more recently, they have been applied to other areas including big data analytics. To understand the advantages and limitations such systems impose on software architects, I migrated an existing big data analytics infrastructure from a software architecture that required lots of work from its developers to deploy and maintain to the new software architecture provided by container-orchestration systems. My results show that scalability is increased, maintenance costs are reduced, and reliability is easier to achieve.