Comparison of Java Microservice Frameworks: Spring Boot, Spring Cloud, Vert.x, and Other Lightweight Options

Spring Boot is easy to set up and benefits from the mature Spring ecosystem, but a simple Spring Boot application already consumes around 304 MB of memory at idle and up to 1.5 GB with CPU usage exceeding 300% under a 30‑second load test.

Using Spring Cloud adds a service‑discovery server, optional gateway, configuration center, distributed tracing, and cluster monitoring, which further increase resource requirements and operational complexity.

Vert.x, a reactive toolkit for the JVM, offers a much lighter footprint: an idle memory usage of about 65 MB, rising to 139 MB with only 2.1% CPU during the same load test, and a startup time under one second.

Vert.x can be used independently or combined with Spring Boot, and powers frameworks such as Apache ServiceComb; it also has an open‑source gateway (VX‑API‑Gateway) supporting multiple languages.

Other lightweight Java microservice frameworks are summarized as follows:

SparkJava : jar ~10 MB, memory 30‑60 MB, performance comparable to Spring Boot.

Micronaut : supports Java, Groovy, Kotlin, offers fast startup and low memory usage, multi‑language, dependency injection, cloud‑native features, but is very new (v1.0.0).

Javalin : extremely easy to adopt, flexible for sync/async, jar 4‑5 MB, about 2 000 lines of source, embedded Jetty server.

Quarkus : fast startup, jar ~10 MB, but documentation is limited.

Overall, Spring Boot requires at least 1 GB of memory for a minimal service, while Vert.x can run many services on a 1‑core, 2 GB server, making it suitable for small projects and modules.