12 Cloud and services

This chapter traces the shift from static, manually provisioned infrastructure to dynamic, elastic cloud environments and argues for precise, shared definitions of cloud, cloud-native, serverless, and services. It presents elasticity—acquiring and releasing resources on demand—as the foundation for scalability and reliability, contrasting earlier proactive, predictive operations with reactive, self-regulating systems. The aim is to provide concise mental models that align with industry practice and improve communication.

Cloud computing is defined by a clear separation between resource consumers and providers and by on-demand acquisition and release of resources. A cloud-native application is scalable and reliable by construction, leveraging platform primitives such as supervisors, autoscalers, and load balancers; lifting and shifting alone does not make an application cloud-native, and specific technologies like containers are not defining characteristics. Serverless computing acquires resources reactively after an event arrives, requiring the system to infer resources per request, and leads to cold and hot paths depending on whether resource acquisition lies on the critical path of request processing.

Reframing microservices as services, the chapter defines a service as a contract between a component and the rest of the system, emphasizing a logical view over the physical set of implementing components. A recommendation service example shows how a stable contract can be realized by a dynamic, redundant assemblage of processes, data stores, load balancers, and autoscalers that scale and fail independently while preserving the same external behavior. The chapter closes by stressing that shared, accurate mental models reduce misalignment and improve engineering and collaboration, encouraging readers to keep probing concepts until they are truly understood.