Azure jaz: Automatic memory and startup optimization for Java in the cloud

Microsoft presents jaz, a JVM launcher that automatically configures and optimizes the Java machine in the Azure cloud. The launcher starts with the container, analyzes its environment and selects the appropriate configuration for the Java VM. The tool promises cloud-native optimized use of memory and faster startup times without the need for manual configuration.

Microsoft's announcement points out that many developers are put off by the complicated configuration of the JVM for the cloud, with the result that thirty percent completely dispense with manual intervention (no further evidence for the figure). Instead, they choose the standard configuration of the Hotspot JVM from Oracle. However, the default settings are conservative and broadly designed and not ideally adapted to the specific environment.

For the Azure cloud, jaz takes over the configuration of the heap size and the garbage strategy, among other things. The tool uses AppsCDS and, in the future, Project Leyden for optimization. According to the announcement, the first deployment of a new Java project in Azure should also be accelerated. All of this should reduce cloud costs.

The tool is compatible with Azure Container Apps, Kubernetes Service, Red Hat OpenShift and Azure VMs and is suitable for Spring Boot, Quarkus or Micronaut projects.

Instead of a manual configuration such as

$ JAVA_OPTS="-XX:... several JVM tuning flags"$ java $JAVA_OPTS -jar myapp.jar"

is just a call of

$ jaz -jar myapp.jar

is required.

By invitation only

In the future, Microsoft plans to integrate telemetry with Prometheus. An invitation (private preview) is currently still required to try out jaz; interested parties can register via a web form.

(who)