Decision-Making Support to Auto-scale Smart City Platform Infrastructures

Smart city platforms support application development and deployment and typically rely on a robust, scalable underlying Information Technology infrastructure composed of cloud platforms, containers, virtual machines, storage, and other services. Such a runtime infrastructure must deal with the highly dynamic workload of the different applications, with simultaneous access from multiple users and sometimes working with many interconnected devices and systems. This scenario requires auto-scaling mechanisms that automatically and timely add or remove cloud resources in response to dynamic variations in workload. This paper introduces a decision-making mechanism that analyzes the monitored state of a smart city platform and its underlying infrastructure at runtime to decide whether auto-scaling is needed. The performance of the decision-making mechanism has been evaluated upon the computational environment that supports a platform for developing real-world smart city applications.