How to Speed Up Spring Boot Startup: 15 Optimizations Tested

This post reproduces the "How fast is Spring?" demo from Spring One 2018 and runs a custom JMH benchmark on a minimal Spring Boot 2.1.0.RELEASE application using OpenJDK 11. The baseline startup time is 2.938 ± 0.287 s/op.

1. FluxBaseline

A simple Reactive Web controller is created with Spring Initializr. The benchmark result is 2.938 ± 0.287 s/op.

@SpringBootApplication
@RestController
public class DemoApplication {
    @GetMapping("/")
    public String home() { return "Hello"; }
    public static void main(String[] args) {
        SpringApplication.run(DemoApplication.class, args);
    }
}

2. WebMVC

Replacing the Reactive controller with a traditional WebMVC controller (Tomcat) yields 3.281 ± 0.342 s/op, slightly slower than the Flux baseline.

3. spring-context-indexer

Adding spring-context-indexer creates a component index. Benchmark: 3.063 ± 0.102 s/op, a small slowdown.

4. Lazy Initialization

Enabling lazy bean initialization reduces startup to 2.844 ± 0.129 s/op.

5. NoVerify

Running the JVM with -noverify improves the time to 2.582 ± 0.060 s/op.

6. TieredStopAtLevel

Using -XX:TieredStopAtLevel=1 dramatically speeds up launch to 1.980 ± 0.037 s/op.

7. Spring Config Location

Specifying -Dspring.config.location=classpath:/application.properties slows it to 3.026 ± 0.139 s/op.

8. Disable JMX

Disabling JMX with -Dspring.jmx.enabled=false yields 2.877 ± 0.097 s/op.

9. Remove Logback

Excluding Logback reduces startup to 2.904 ± 0.096 s/op.

10. Remove Jackson

Excluding the Jackson starter brings the time down to 2.789 ± 0.093 s/op.

11. Remove HibernateValidator

Excluding Hibernate Validator results in 2.857 ± 0.084 s/op.

12. AppCDS

Enabling Application Class Data Sharing (AppCDS) gives 2.957 ± 0.079 s/op, showing no improvement because Spring Boot fat JARs are not covered by CDS.

13. Thin Launcher (Exploded)

Packaging with the Spring Boot Thin Launcher (exploded layout) reduces launch time to 2.476 ± 0.091 s/op.

14. Thin Launcher + AppCDS

Combining Thin Launcher with AppCDS yields a significant gain: 1.535 ± 0.036 s/op.

15. All Optimizations Applied

Applying every previous tweak together results in a sub‑second startup: 0.801 ± 0.037 s/op.

The final benchmark table summarises all cases:

Benchmark                                 Mode  Cnt   Score   Error  Units
MyBenchmark.case01_FluxBaseline           ss    10   2.938 ± 0.287  s/op
MyBenchmark.case02_Web                    ss    10   3.281 ± 0.342  s/op
MyBenchmark.case03_WithContextIndexer      ss    10   3.063 ± 0.102  s/op
MyBenchmark.case04_WithLazyInit           ss    10   2.844 ± 0.129  s/op
MyBenchmark.case05_WithNoVerifyOption     ss    10   2.582 ± 0.060  s/op
MyBenchmark.case06_WithTieredStopAtLevel1  ss    10   1.980 ± 0.037  s/op
MyBenchmark.case07_WithSpringConfigLocation ss   10   3.026 ± 0.139  s/op
MyBenchmark.case08_WithJmxDisabledOption  ss    10   2.877 ± 0.097  s/op
MyBenchmark.case09_WithoutLogback         ss    10   2.904 ± 0.096  s/op
MyBenchmark.case10_WithoutJackson          ss    10   2.789 ± 0.093  s/op
MyBenchmark.case11_WithoutHibernateValidator ss 10   2.857 ± 0.084  s/op
MyBenchmark.case12_WithAppCds             ss    10   2.957 ± 0.079  s/op
MyBenchmark.case13_Exploded               ss    10   2.476 ± 0.091  s/op
MyBenchmark.case14_ExplodedWithAppCds     ss    10   1.535 ± 0.036  s/op
MyBenchmark.case15_AllApplied              ss    10   0.801 ± 0.037  s/op

These experiments demonstrate how various JVM flags, Spring Boot configuration choices, and dependency exclusions can collectively shrink Spring Boot startup time from nearly three seconds to under one second.