Adaptive computing systems automatically monitor their behavior and dynamically adjust their own configuration parameters—or knobs—to ensure that user goals are met despite unpredictable external disturbances to the system. A major limitation of prior adaptation frameworks is that their internal adaptation logic is implemented for a specific, narrow set of goals and knobs, which impedes the development of complex adaptive systems that must meet different goals using different sets of knobs for different deployments, or even change goals during one deployment.

To overcome this limitation we propose GOAL, an adaptation framework distinguished by its virtualized adaptation logic implemented independently of any specific goals or knobs. GOAL supports this logic with a programming interface allowing users to define and manipulate a wide range of goals and knobs within a running program. We demonstrate GOAL’s benefits by using it re-implement seven different adaptive systems from the literature, each of which has a different set of goals and knobs. We show GOAL’s general approach meets goals as well as prior approaches designed for specific goals and knobs. In dynamic scenarios where the goals and knobs are modified at runtime, GOAL achieves 93.7% of optimal (oracle) performance while providing a 1.69\times× performance advantage over existing frameworks that cannot perform such dynamic modification.