logo
Dates

Author


Conferences

Tags

Sort by:  

Conference:  ContainerCon 2022
Authors: Chen Wang, Huamin Chen
2022-06-21

Do you want to help combat climate change? Are you concerned with the electricity cost of your Kubernetes clusters? Then join our efforts to manage energy efficiency on Kubernetes clusters. Currently, the energy consumption metrics are only available at node levels. There is no way to obtain container-level energy consumption. Autoscalers and schedulers really need pod-level metrics data in order to obtain energy savings from resizing or migrating containers. We present Kubernetes-based Efficient Power Level Exporter (Kepler) and its integration with Kubernetes. By leveraging eBPF programs, Kepler probes per container energy consumption related system counters and exports them as metrics. These metrics help end users observe their containers’ energy consumption and allow cluster admins to make intelligent decisions on achieving energy conservation goals. We demonstrate that the Kepler can be easily integrated into Prometheus and the existing dashboard.
Authors: Ying-Feng Hsu
2021-10-14

Many K8s extensions have been focused on large scale container computation. But, how to strike a balance between energy efficiency and service performance for container operations due to the continuous growth of IoT devices and edge computing systems? The current K8s does not provide container orchestration from the perspective of data center power reduction. This talk presents a Workload Allocation Optimizer (WAO) based on the K8s architecture. WAO uses ML to predict the power increasing of workloads and introduces a scoring plugin to the K8s scheduler framework for Node selection. WAO-load balancer enables Pods to Nodes assignment with optimal power consumption. This talk gives you details on how power saving can be realized for cloud-edge computing systems. Instead of using the virtual environment, we demonstrate the proposed WAO in a real edge data center with 200+ servers and show you how WAO manipulates the tradeoff between service performance and data center power saving.