Building Resource Auto-scaler with Functional-Link Neural Network and Adaptive Bacterial Foraging Optimization.

In this paper, we present a novel intelligent proactive auto-scaling solution for cloud resource provisioning systems. The solution composes of an improvement variant of functional-link neural network and adaptive bacterial foraging optimization with life-cycle and social learning for proactive resource utilization forecasting as a part of our auto-scaler module. We also propose several mechanisms for processing simultaneously different resource metrics for the system. This enables our auto-scaler to explore hidden relationships between various metrics and thus help make more realistic for scaling decisions. In our system, a decision module is developed based on the cloud Service-Level Agreement (SLA) violation evaluation. We use Google trace dataset to evaluate the proposed solution well as the decision module introduced in this work. The gained experiment results demonstrate that our system is feasible to work in real situations with good performance.