How to Optimize JVM Memory Settings for Spring Boot in Limited Environments

Background

During a migration from Spring Boot to Spring Cloud, the number of deployed services grew and the server began to run out of memory. Most services were started without any JVM memory flags, causing each instance to consume 1.5‑2 GB of RAM by default.

JVM Default Memory Settings

If no -Xms or -Xmx options are supplied, the JVM uses its own default strategy, which differs for large‑memory and small‑memory environments.

In a typical 4 GB machine, the defaults are:

The maximum heap is 1/4 of physical memory, and the initial heap is 1/64 of physical memory. If the application exceeds the maximum, an OutOfMemoryError is thrown.

When the physical memory is less than 192 MB (e.g., on a mobile device), the defaults change:

In this case the maximum heap becomes 1/2 of physical memory, while the initial heap remains 1/64 but not less than 8 MB.

Practical Configuration

The simplest way to avoid memory pressure is to set explicit values when launching the jar: java -Xms64m -Xmx128m -jar xxx.jar Note that the memory flags must appear before the -jar option; otherwise they are ignored.

Inspecting JVM Settings

View System Default Memory Settings

On Linux:

java -XX:+PrintFlagsFinal -version | grep HeapSize

On Windows:

java -XX:+PrintFlagsFinal -version | findstr HeapSize

List Running JVM Processes

jps

Typical output shows process IDs and the main class name.

Check GC Statistics for a Specific Process

jstat -gc 51972 5000

The columns include survivor space capacities (S0C, S1C), Eden space (EC, EU), old generation (OC, OU), and GC counts/times (YGC, YGCT, FGC, FGCT, GCT).

Examine Heap Usage with jmap

jmap -heap 10471

The command prints configuration values such as MaxHeapSize, NewSize, OldSize, and detailed usage of young and old generations.

Conclusion

Neglected JVM memory defaults can quickly exhaust server resources when many Spring Boot services run side‑by‑side. By explicitly configuring -Xms and -Xmx, and by using Java’s built‑in inspection tools ( java -XX, jps, jstat, jmap), developers gain fine‑grained control over heap allocation and can prevent out‑of‑memory crashes.