Container Load Placement for Deep Learning Application using Whale Optimization
The deployment and scaling of deep learning applications in distributed computing environments pose significant challenges, particularly in the context of containerized virtualization. Efficient placement and management of Docker containers are critical to optimizing resource utilization, minimizing latency, and ensuring the scalability of deep learning models across clusters of machines. In this paper, we will compare five methods of container placement that are implemented within a scheduling method named Differentiate Quality of Experience Scheduling (DQoES). The container placement methods to be compared include the default Docker Swarm container placement, Discrete Whale Optimization Container Placement (DOWCP), proposed whale optimization, a proposed hybrid with DWOCP, and a proposed hybrid with proposed whale optimization. Based on the experimental results, the method that demonstrates better performance than both the default Docker Swarm container placement and DWOCP is the proposed hybrid with proposed whale optimization.