Thoras ingests a targeted set of JVM runtime metrics exposed through JMX into
its forecasting model to drive accurate CPU and memory predictions for Java
workloads. Based on these predictions, Thoras rightsizes both the pod resource
requests and dynamically updates JVM runtime parameters (such as heap size) to
ensure optimal performance. For detailed guidance on configuring JVM-specific
scaling strategies, reach out to your Thoras point-of-contact.