Neural Adaptive Admission Control Framework: SLA-driven action termination for real-time application service management

Although most modern cloud-based enterprise systems, or operating systems, do not commonly allow configurable/automatic termination of processes, tasks or actions, it is common practice for systems administrators to manually terminate, or stop, tasks or actions at any level of the system. The paper investigates the potential of automatic adaptive control with action termination as a method for adapting the system to more appropriate conditions in environments with established goals for both system’s performance and economics. A machine-learning driven control mechanism, employing neural networks, is derived and applied within data-intensive systems. Control policies that have been designed following this approach are evaluated under different load patterns and service level requirements. The experimental results demonstrate performance characteristics and benefits as well as implications of termination control when applied to different action types with distinct run-time characteristics. An automatic termination approach may be eminently suitable for systems with harsh execution time Service Level Agreements, or systems running under conditions of hard pressure on power supply or other constraints. The proposed control mechanisms can be combined with other available toolkits to support deployment of autonomous controllers in high-dimensional enterprise information systems