Abstract
The exponential surge in the usage of container based applications hosted on datacenters has resulted in a significant increase in energy consumption. This highlights the necessity for optimized container orchestration systems, thus making it important for infrastructures that provide Container-as-a-Service (CaaS) to have efficient scheduling, rescheduling, resource provisioning and container networking. Our work proposes various methods for optimizing constraints associated with container orchestration which include resource utilization, network data footprint, and energy efficiency. The aforementioned constraints are optimized by two new approaches. The first one is a hybrid algorithm that utilizes the concepts of genetic and bin-packing algorithms for better resource utilization. The second approach is based on a dynamic programming paradigm to reduce network data footprint. Furthermore, for improving energy efficiency, we have employed Dynamic Voltage and Frequency Scaling (DVFS). By putting together these approaches to optimize container orchestration, we noticed an overall improvement which has led to a decrease in energy consumption by 22.88% and the reduction in network usage by 16.04%
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call