2021· International Journal of Machine Learning and Predictive Analytics· 0 citations
TL;DR
Reinforcement learning-based adaptive resource management framework is proposed that enables cloud systems to autonomously learn optimal resource allocation policies through continuous interaction with the environment and significantly outperforms static and reactive baseline strategies in terms of resource utilization efficiency and response time stability.
Abstract
Cloud software systems operate under highly dynamic and unpredictable workloads, requiring efficient and adaptive resource management strategies to maintain performance, reliability, and cost efficiency. Traditional rule-based and heuristic resource allocation approaches often fail to respond optimally to rapid workload fluctuations and complex system interactions. This paper proposes a reinforcement learning-based adaptive resource management framework that enables cloud systems to autonomously learn optimal resource allocation policies through continuous interaction with the environment. By modeling cloud resource management as a sequential decision-making problem, the framework leverages reinforcement learning algorithms such as Q-learning, Deep Q-Networks (DQN), and policy-gradient methods to dynamically adjust computing resources including CPU, memory, and virtual machine instances. The proposed approach aims to optimize multiple objectives such as performance, cost, and service-level agreement (SLA) compliance. Experimental evaluation using simulated and real-world cloud workloads demonstrates that reinforcement learning significantly outperforms static and reactive baseline strategies in terms of resource utilization efficiency and response time stability. The results highlight the potential of reinforcement learning to enable intelligent, self-adaptive cloud resource management systems.
The rapid growth of cloud computing has significantly increased the demand for efficient resource management techniques capable of supporting large-scale multi-tenant cloud environments. As cloud infrastructures continue to expand, managing heterogeneous computing resources while ensuring scalability, optimal resource utilization, Quality of Service (QoS), Service Level Agreement (SLA) compliance, energy efficiency and reduced operational costs has become increasingly challenging. Existing resource management approaches often suffer from poor scalability, high computational overhead, inefficient workload distribution and limited adaptability to dynamic workload variations. This study presents an efficient algorithm for large-scale resource management in a multi-tenant cloud environment. The proposed framework integrates intelligent resource scheduling, workload balancing and adaptive virtual machine allocation to optimize resource utilization while satisfying multiple performance objectives. An object-oriented system development methodology was employed to design the framework, while a Deep Reinforcement Learning (DRL)-based optimization algorithm was implemented to enable autonomous decision-making through continuous learning from workload patterns, resource states and environmental feedback. The proposed algorithm efficiently allocates and manages cloud resources across multiple tenants, minimizing resource contention, improving system throughput, reducing response time and energy consumption and enhancing overall cloud performance. Experimental evaluation demonstrates that the proposed approach provides a scalable, adaptive and computationally efficient solution for large-scale resource management in modern multi-tenant cloud environments.
Onwuegbuchulem Gift., Bennett E.O., M. D. et al.· International journal of re...· 0 citations
This paper proposes an innovative Deep Reinforcement Learning-based Intelligent Task Scheduling Framework (DRITS) designed to optimize task allocation and resource utilization in cloud distributed systems and establishes DRL-based intelligent scheduling as a promising solution for next-generation cloud computing infrastructure management.
Tileemat Ashour Aletiri· مجلة العلوم الشاملة· 0 citations
Cloud computing has evolved significantly, with multi-cloud environments becoming popular for improving scalability, reliability, fault tolerance, and cost efficiency. However, resource scheduling in multi-cloud systems remains challenging due to heterogeneous infrastructures, dynamic workloads, varying pricing models, network latency, and SLA requirements. Traditional scheduling methods such as Round Robin and FCFS often fail to adapt effectively to these complexities. Artificial Intelligence (AI) offers an advanced solution through intelligent resource scheduling. By leveraging machine learning, reinforcement learning, and predictive analytics, AI-based schedulers can forecast workloads, optimize resource allocation, and make autonomous scheduling decisions. This research proposes an AI-driven resource scheduling framework that integrates workload prediction, resource classification, intelligent scheduling, and continuous feedback mechanisms. The framework aims to optimize multiple objectives, including cost reduction, execution efficiency, energy consumption, and SLA compliance. Performance evaluation using metrics such as resource utilization, makespan, response time, throughput, and energy consumption demonstrates that AI-assisted scheduling outperforms traditional approaches. The results indicate improved resource utilization, better workload balancing, reduced operational costs, and enhanced service quality, highlighting the potential of AI-driven scheduling for next-generation multi-cloud resource management systems.
Michael Anderson· International Journal of App...· 0 citations
Sensitivity and ablation studies confirm stable learning and controllable latency-cost trade-offs, demonstrating that lightweight RL can effectively deliver cost-efficient, adaptive autoscaling in hybrid cloud environments.
Bekzat Kobei, N. Seilova, Zarina A. Kashaganova· AI@DTESI· 0 citations
An AI-enabled dynamic task scheduling framework based on Deep Reinforcement Learning (DRL) with a Deep Q-Network (DQN) model to dynamically assign tasks to virtual machines and learn the best scheduling policies by continuously interacting with the cloud environment based on system parameters such as resource availability, task queue length, and virtual machine load is introduced.
Karnam Sreenu, G. Prasadu, K. Premnadh et al.· VFAST Transactions on Softwa...· 0 citations
The rapid growth of Big Data, IoT, cloud computing, edge intelligence, and AI has increased the demand for scalable and efficient analytical infrastructures. Traditional distributed computing systems often rely on fixed resource allocation and execution strategies, leading to performance issues, resource underutilization, higher latency, and limited scalability under dynamic workloads. To address these challenges, self-adaptive distributed computing models enable systems to autonomously monitor, analyze, and optimize operations in real time. The proposed framework integrates adaptive resource management, intelligent workload scheduling, dynamic task migration, predictive analytics, and machine learning-based optimization to improve computational efficiency and responsiveness. The architecture includes monitoring layers, decision engines, adaptation controllers, distributed resource managers, and analytics execution frameworks. Machine learning techniques such as reinforcement learning, deep neural networks, and predictive modeling support proactive adaptation by forecasting workload demands and optimizing scheduling decisions. Experimental results demonstrate significant improvements in resource utilization, throughput, scalability, fault tolerance, and execution time compared to traditional approaches. Self-healing capabilities further enhance resilience against node failures and network disruptions. Overall, self-adaptive distributed computing provides a scalable, intelligent, and resilient foundation for next-generation high-performance analytics and data-intensive applications.
John Peterson, L. Martínez· International Journal of App...· 0 citations