Autonomic Cloud Computing: Research Perspective
Sukhpal Singh Gill
TL;DR
The paper examines the tension between SLA guarantees and QoS in large-scale, dynamic cloud environments and argues for autonomic cloud computing as a viable path to maintain performance without human intervention. It surveys SLA and QoS concepts, analyzes SLA deviations, and demonstrates through CloudSim-based validation that autonomic management can improve QoS and reduce SLA violations. A central contribution is a conceptual AI-driven autonomic cloud model built around the MAP(E)-K loop, with a focus on self-* properties (configuration, healing, optimization, protection) and an architecture that integrates AI for adaptive resource management. The discussion identifies future directions in security/privacy, Edge AI, and energy efficiency, highlighting potential practical benefits such as cost savings, resilience, and reduced environmental impact, alongside open research questions for AI-enabled autonomic systems.
Abstract
As the cloud infrastructure grows, it becomes more challenging to manage resources in such a massive, diverse, and distributed setting, despite the fact that cloud computing provides computational capabilities on-demand. Due to resource variability and unpredictability, resource allocation issues arise in a cloud setting. A Quality of Service (QoS) based autonomic resource management strategy automates resource management, delivering trustworthy, dependable, and cost-effective cloud services that efficiently execute workloads. Autonomic cloud computing aims to understand how computing systems may autonomously accomplish user-specified "control" objectives without the need for an administrator and without violating the Service Level Agreement (SLA) in a dynamic cloud computing environments. This chapter presents a research perspective and analysis on autonomic resource allocation in cloud computing based on the last decade of conducted research with a focus on QoS and SLA-aware autonomic resource management. This study delves into the current state of autonomic resource management in the cloud and introduces a conceptual model for Artificial Intelligence (AI)-driven autonomic cloud computing. This model aims to optimise server load distribution and energy consumption, thus enhancing cost savings and environmental impact. Finally, it highlights key next-generation research directions.