The Java-based Teradata OTF query processing is memory and CPU intensive. All OTF queries run within a single, multi-threaded JVM (Java OTF server) on a node.
The memory available to the OTF queries is restricted by the JVM max memory limit which is set to 15GB.