Virtual Machine Allocation in Cloud Computing Using Reinforcement Learning: DDPG

Virtualization and cloud computing can be used together in new ways to help cloud-based data centres (DCs) make better use of their limited resources and use less power. A big problem with cloud data centres is that they use a lot of power. This is bad for the environment and raises the costs for pe...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2023 4th International Conference on Computation, Automation and Knowledge Management (ICCAKM) s. 1 - 8
Hlavní autoři: Panesar, Gurpreet Singh, Chadha, Raman
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 12.12.2023
Témata:
Ant Colony System
Cloud computing
Cloud Data Centers
Data centers
DDPG
Dynamic Workloads
Energy Efficiency
Genetic Algorithm
Genetic algorithms
Particle Swarm Optimization
Performance Evaluation
Quality of service
Reinforcement learning
Resource management
Resource Utilization
Virtual Machine Allocation
Virtual machining
Virtualization
VM Consolidation
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Virtualization and cloud computing can be used together in new ways to help cloud-based data centres (DCs) make better use of their limited resources and use less power. A big problem with cloud data centres is that they use a lot of power. This is bad for the environment and raises the costs for people who use the cloud. VM consolidation is something that some people think could help solve this problem. Moving VMs to different real hosts is part of this method. This spreads them out and makes them use less power. Using a method called Deep Deterministic Policy Gradient (DDPG) for reinforcement learning to group virtual machines (VMs) together is a new way to make cloud data centres work better. When you use DDPG to group virtual machines together in a cloud data centre, you save money and time. Common methods like the Genetic Algorithm, Particle Swarm Optimisation, and the Ant Colony System are not the same as the work that is being offered. This method and the VM transfer work better in simulations than popular ones like the genetic algorithm (GA), particle swarm optimisation (PSO), and adaptive control system (ACS). If you do things the way that is suggested, you will use 72% less material and 68 KWh of energy. The tests also show that the proposed way cuts down on the number of running physical machines (PMs) by a large amount.
DOI:10.1109/ICCAKM58659.2023.10449636