Towards a Sustainable Cloud-Native Platform
summary
The talk explains how to build a more environmentally sustainable cloud-native platform by optimizing at OS/node, cluster and system levels using existing CNCF technologies. It covers energy-aware node and CPU controls (eco profiles, CPU management, frequency control), intelligent workload placement, dynamic scaling and AI/ML-based forecasting, plus techniques to reduce network footprint and scale clusters based on real-time resource demand, and highlights gaps in the Kubernetes ecosystem with real-world examples.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 2:16Greenhouse gas scopes
- 4:21Energy impact measurement
- 7:09Sustainability observability
- 8:34Efficiency data analysis
- 9:45Optimizing with actions
- 10:48Node and cluster optimization
- 13:16System and domain optimization
- 16:04CPU power states
- 17:27Autoscaling strategies
- 18:28CPU profiling for apps
- 19:51Kepler energy measurement
- 21:14Open-source sustainability architecture
- 22:17Carbon Aware Scheduling
- 23:00Java and Quarkus optimizations
- 24:50Scopes 1, 2 and 3
- 25:36Financial and energy savings
- 26:59Q&A
related talks
12mKeynote: Cloud Native Sustainability - How You Can Help Right Now!
07mKeynote: Exploring the Frontiers of Cloud Native
47mHow Kubernetes optimisation can combat climate change
27mCloud Native Landscape for Node.js Developers
09mKeynote: Build an Open Source Distributed Cloud Native World
48mCloud Native Applications on Kubernetes: a DevOps approach
23mMaking Cloud Native Computing Universal and Sustainable
41mTales of the mythical Cloud-Native Platform