Modern applications as well as their environments should enable scalability when the workload on them increases. Applications ideally are able to scale out horizontally, rather than just vertically. The difference is that scaling horizontally aims to adds more individual, self-contained nodes to the software whereas scaling vertically increases the resources on single nodes or processes. However, scaling vertically is limited, since resources on physical nodes cannot be increased infinitely.

12-factor applications describe the procedure of adding concurrency to the software with adding more self-contained, shared-nothing processes. Workloads should be distributable within several physical hosts, by increasing the number of processes. The processes represent the request or worker threads who handle the system's workload.

This approach shows the necessity of implementing stateless application in a shared-nothing manner. Containers that run stateless Java enterprise applications enable the system to scale out. Kubernetes managed scalability in deployments via managing the number of replicas.

The bottleneck of enterprise applications, however, is typically not the application instances rather than central databases. Chapter 8, Microservices and System Architecture and Chapter 9, Monitoring, Performance, and Logging cover the topics of scalability in distributed systems as well as performance in Java EE projects in general.