In a given JVM, the Corticon Server and its Decision Services occupy the following amounts of physical memory:
State of Corticon Server
RAM required
Basic Corticon Server overhead with no Decision Services (excludes memory footprint of the JVM which varies by JDK version and platform)
25 MB
Load a single Decision Service from the Deployment Descriptor or addDecisionService() method API.
~ 5 MB (this is the overhead for the Trade Allocation sample application with seven (7) Rulesheets, twenty-four (24) rules, five (5) associated entities)
Working memory to handle a single CorticonRequest
~ 1 MB (this is the overhead for the Trade Allocation sample application with seven (7) Rulesheets, twenty-four (24) rules, five (5) associated entities (steady-state usage)).
You may reduce the amount of memory required in a large system by dynamically loading and unloading specific Decision Services. This is especially relevant in resource-constrained handheld or laptop scenarios where only a single business transaction occurs at a time. After the first Decision Service is invoked, it is unloaded by the application and the second Decision Service is loaded (and so on). While this will be slower than having all Decision Services always loaded, it can address tight, memory-constrained environments. A compromise alternative would only dynamically load/unload infrequently used Decision Services.
