活动名称：全球 CNCF 可持续发展周 - 中国Meetup

目的： CNCF 环境可持续发展技术咨询小组 (TAG ENV) 组织了一系列Meetup。这是全球云原生可持续发展周的一部分，旨在解决云原生行业和开源社区中环境可持续性的主题。

地点：北京、上海，在线视频会议

时间： 2023年11月4日

强调：

1. 我们举办了 China Meetup，介绍了云原生可持续发展的前景、目标和项目（ https://tag-env-sustainability.cncf.io/landscape/ ），包括介绍云原生环境中的可持续计算、可观测工具与其他工具的集成多个技术领域视角，例如平台适配、自动化、优化和调度、碳感知框架。

2. 4小时的Meetup共有4,400+在线观看人次，加上北京和上海会场15人亲自参与。

3. 北京、上海以及在线视频会议的照片。

3. IBM和 Intel的总共 5 场演讲，其中有很多 Q/A 和良好的评论。英特尔在 Meetup 中发表了两场演讲。

4. Intel的演讲

标题：云原生 AI 管道的碳感知框架

作者：Lu Ken，英特尔软件经理

摘要：

基于容器、命名空间和集群的碳足迹指标，构建闭环的碳感知云原生编排和扩展策略，并通过碳算子应用策略，实现可持续的云原生AI管道的节能减排。

Kubernetes 设备插件将平台的异构计算引擎和加速器透明传输到云原生微服务或功能服务，从而无需修改即可优化工作负载。

为行业提供跨硬件参考平台、跨编排架构、跨云边缘部署的标准化云原生碳足迹公式和计算框架，促进碳信用交易市场快速发展，提高全行业的碳足迹意识行业。

视频录制： 2:57~3:46

标题：Kepler 平台验证框架和准确性检查

作者：任杰，英特尔高级工程师

摘要：

Kepler（Kubernetes Efficient Power Level Exporter）是一个 CNCF Sandbox 项目，它使用eBPF来探测云原生容器的与能源相关的系统统计数据和性能事件计数器，并将它们导出为 Prometheus 指标。它支持功率比建模和功率估计建模。在本次演讲中，我们介绍了Kepler中最新上游的平台验证框架，对项目的测试框架进行有益的补充，使开发人员和最终用户能够在特定的硬件平台上验证项目的功能。

供应商特定的验证案例可以集成到框架中。我们可以检查验证的测试报告，看看开普勒是否很好地支持供应商的平台，以评估和识别差距和限制，特别是在功率建模的准确性方面。

平台验证框架是完全自动化的，并且与运行器系统无关，我们还可以在手动测试用例执行和演示使用中利用它。

视频录制： 0:59~1:30

5. 视频记录文件：

哔哩哔哩： https://www.bilibili.com/video/BV1BN4y1S7kt/ ?vd_source=19c882661707a5e827accf928b20054e
YouTube：https://youtu.be/ZcIJvRmVt1Q

感谢红帽、IBM、英特尔赞助和主办本次聚会！

========================================================

Event Title: Global CNCF Sustainability Week - China Meetup

Purpose: A series of Meetups have been organized by CNCF Environmental Sustainability Technical Advisory Group (TAG ENV). This is part of effort of the Global Week of Cloud Native Sustainability, to address the emerging topic of environmental sustainability in the cloud native industry and open source community.

Where: Beijing and Shanghai, with online video conference

When: Nov. 4th, 2023

Highlights:

1. We hosted China Meetup to introduce Cluod Native Sustainbility landscape, goals and projects (https://tag-env-sustainability.cncf.io/landscape/), including introduction of sustainable computing in a cloud-native environment, observability tooling integration with other multiple technological area perspectives such as platform adaptation, automation, optimization, and scheduling, carbon-aware framework.

2. Total 4,400+ online view during 4 hours’ Meetup, plus 15 people participated in person in Beijing and Shanghai venue.

3. Photos from Beijing, Shanghai, and online video conference.

4. Total 5 presentations from Red Hat, IBM and Intel, with a lot of Q/A and good comments. Intel contributed two presentations in Meetup.

Title: Carbon-Aware Framework for Cloud Native AI Pipeline

By: Lu Ken, Intel Software Manager

Abstract:

Based on the carbon footprint indicators of containers, namespaces and clusters, build a closed-loop carbon-aware cloud-native orchestration and expansion strategy, and apply strategies through carbon operators to achieve energy conservation and emission reduction in a sustainable Cloud-Native AI pipeline.

Kubernetes device plugins transparently transmit the platform's heterogeneous computing engines and accelerators to cloud-native microservices or function services, allowing workloads to be optimized without modification.

Provide the industry with standardized cloud-native carbon footprint formulas and calculation frameworks across hardware reference platforms, cross-orchestration architectures, and cross-cloud edge deployments to facilitate the rapid development of the carbon credit trading market to improve the carbon footprint awareness of the entire industry.

Video record: 2:57~3:46

Title: Kepler Platform Validation Framework and accuracy check

By: Ren Jie, Intel Sr. Engineer

Abstract:

Kepler (Kubernetes Efficient Power Level Exporter) is a CNCF Sandbox project, it uses eBPF to probe energy-related system statistics and performance event counters for cloud native containers and exports them as Prometheus metrics. It supports Power Ratio Modeling and Power Estimation Modeling. In this talk, we introduced the newly upstreamed platform validation framework in Kepler to make beneficial supplement to the project’s test framework, enable the developers and end users to validate the project features on specific hardware platforms.

Vendor specific validation cases could be integrated into the framework. We could check test reports of the validation to see if the vendors’ platforms are well supported in Kepler, to evaluate and identify the gap and limitation, especially on the accuracy of the power modeling.

The platform validation framework is fully automated and runner system agnostic, we can also leverage it in manual test cases execution and demo use.

Video record: 0:59~1:30

5. Video record files:

Bilibili: https://www.bilibili.com/video/BV1BN4y1S7kt/?vd_source=19c882661707a5e827accf928b20054e

Youtube: https://youtu.be/ZcIJvRmVt1Q

Thanks for Red Hat, IBM, Intel for their sponsoring and hosting this Meetup!