Efficient Algorithm for Multi-Objective Dynamic Resource Allocation in Cloud Computing

Abstract

Cloud computing environments depend heavily on efficient Dynamic Resource Allocation (DRA) mechanisms to ensure optimal utilization of computational resources while maintaining low operational cost, reduced energy consumption, and acceptable Quality of Service (QoS) under continuously fluctuating workloads. However, many existing resource allocation techniques in cloud systems are limited by poor adaptability, high computational overhead, inefficient virtual machine migration, and inability to simultaneously optimize multiple conflicting objectives such as throughput, Service Level Agreement (SLA) compliance, and power efficiency. These limitations create the need for a more intelligent, scalable and adaptive resource management framework capable of making real-time allocation decisions in heterogeneous cloud environments. This study therefore presents the design and development of DynamiCloud, a scalable and computationally efficient multi-objective dynamic resource allocation model for cloud computing. The research aimed at developing an efficient algorithm for multi-objective Dynamic Resource Allocation (DRA) in cloud computing. An object-oriented system design methodology was adopted in modeling the proposed framework; while a Deep Reinforcement Learning (DRL)-based optimization algorithm was implemented to enable the system learn optimal VM allocation and reallocation policies from environmental states, reward signals, and workload behavior patterns. The design was implemented using python. Comparing the results of our implementation with the existing tools shows that our objectives were met.