Abstract
Cloud computing has drastically changed the access and management of computational resources in cloud data centers, allowing enterprises to dynamically scale their services in response to increasing demands. Despite these challenges, managing and using these resources effectively remains difficult because of financial issues, pressure on the network, and the need for easy growth and resource management. As businesses move their services to the cloud, it is important to make virtual machine (VM) setup and resource use more efficient. This helps reduce costs and improve system performance.
This dissertation looks at the challenges of automatically assigning resources to virtual machines in cloud networks. The study looks at using the Network Design and Evolution Optimization (NetDEO), which uses swarm intelligence, to improve how VMs are placed in cloud systems. NetDEO’s unique approach allows for the dynamic adjustment of VMs based on real-time demand, reducing network traffic stress, and enhancing resource utilization efficiency in data centers. The algorithm was evaluated by placing virtual machines in different network setups, like FatTree, BCube, and Tree. The evaluation focused on how well it balanced the load of the VMs, reduced traffic jams, and adjusted to changing workloads.
Each network design was evaluated to see how well it manages traffic, grows with demand, manages problems, and controls costs in cloud data centers. Simulation results show that NetDEO significantly outperforms legacy VM allocation methods by eliminating system stress and balancing traffic load across nodes and preventing VM overload. In the FatTree design, known for its efficient and scalable structure, NetDEO evenly spreads the load across all VM servers. This helps prevent slowdowns and ensures smooth data transfer for cloud users. The BCube topology demonstrates high availability, with the algorithm effectively distributing traffic and preventing critical points of failure on the network. On the other hand, the tree topology has major issues with imbalance traffic. It tends to put too much strain on top-level nodes, which can lead to bottle necks on the network. The topology is not ideal for handling large amounts of data cloud data centers.
Key performance metrics such as maximum system traffic stress, average traffic stress, were used to evaluate the algorithm’s effectiveness. The results show that using NetDEO to automate where VMs are placed can help cloud networks to perform well while cutting cost. This is a useful solution for organizations wanting to manage their cloud resources better. This study offers a strong plan for improving cloud infrastructure and shows how AI algorithms can be used to manage resources and balance workloads based on what cloud users need.
This dissertation contributes to the broader field of cloud network management by presenting a scalable, efficient approach to VM allocation. It offers actionable recommendations for organizations seeking to enhance resource allocation cost-effectively while reducing network congestion that could impact service delivery to cloud users. Future research could explore hybrid...