Java applications running on Kubernetes are expected to start quickly, consume less memory, use compact container images, and scale efficiently. Achieving all four is not a single runtime decision. Improvements at the container, framework, compilation, and execution layers affect different measurements and introduce different operational costs.

Using a working Java application, Daniel will demonstrate how open-source projects and tools can improve startup and readiness time, memory footprint, image size, throughput, and workload density. You will examine Jib, Quarkus, Native Image, CRaC, and related technologies as complementary approaches operating at different layers rather than as competing products in a simple benchmark.

The session connects each technique to the Kubernetes workload characteristics that make it useful, including elastic scaling, scale-to-zero, long-running services, and high-density deployments. You will also examine the costs that headline performance numbers can hide, including build complexity, compatibility constraints, checkpoint security, reduced debugging facilities, and steady-state throughput. Where comparisons with other programming languages are useful, the workload and test conditions will be made explicit.

What You Will Learn

• How different techniques affect startup and readiness time, memory footprint, image size, throughput, and workload density

• Where Jib, Quarkus, Native Image, CRaC, and related technologies fit within the Java and Kubernetes stack

• How workload characteristics such as elastic scaling, scale-to-zero, long-running services, and high-density deployments influence optimisation choices

• How to weigh performance improvements against operational costs and constraints

Who Should Attend

Java Developers, Platform Engineers, Software Architects, SREs, and Technical Leads building or operating Java applications on Kubernetes.