Architectural Principles for Cloud-Scale Data Systems

Cloud-scale services exhibit an increasing diversity of workloads and user demands, reflected in a growing range of deployment models by the service consumers. Multitenancy, autoscaling, performance isolation, energy and carbon footprint reduction, and lifecycle-management support are now a requirement rather than a luxury. Architectural trends clearly point in the direction of service-oriented or microservices architectures. However, casuistics about cloud-scale services also evidence a wide range of rarely discussed data-centric aspects of their architectures. Addressing these considerations in a structured manner can help cloud-scale data systems evolve toward appropriate common patterns and paradigms that ease their adoption and operation and constitute sound foundations for training cloud computing practitioners.