AI
TAG Operational Resilience: Mission and Structure
The CNCF TOC restructured all technical advisory groups last year, collapsing prior groups into four or five focused areas. TAG Operational Resilience absorbed observability, management, business continuity, resource optimization, cost efficiency, green initiatives, reliability, and day-two operations. The TAG’s charter: define practices and standards for building, operating, and managing resilient and efficient cloud-native systems. Work is organized as scoped, time-bound initiatives rather than open-ended working groups.
Project Release Guidelines: Helping CNCF Projects Ship Safely
Many sandbox, incubating, and even graduated CNCF projects lack formal guidance on code management, artifact publishing, distribution, and supply chain security. The release guidelines initiative addresses that gap by producing patterns and reference implementations drawn from existing projects. A draft white paper is already available for public comment. Project maintainers can contribute requirements or participate as case studies via the CNCF Slack initiative channel or the TAG operational resilience channel.
Observability Personas: A Shared Definition Across CNCF
Operators, developers, and other observability consumers use metrics, traces, logs, and profiles in different ways. Without a common definition of who each persona is and what signals they rely on, CNCF projects talk past each other and end users struggle to align internal strategies. This new initiative interviews end-user groups and companies to build a shared reference. It also targets reducing alert fatigue by making persona definitions explicit across projects. AI observability adds another persona category the group is already tracking.
Sustainability Month and Green Reviews: Measuring Carbon Footprint
The TAG runs an annual cloud-native sustainability month, with community-organized meetups held in person and virtually across cities including Tokyo and Barcelona. The green reviews project measures the sustainability footprint of CNCF projects using Kepler and related tools. Mario Fahlandt noted a mid-to-long-term goal: make a passing green review a formal requirement for projects moving from sandbox to incubation or graduation, published in the TOC repo alongside existing security and governance reviews.
How to Start or Join an Initiative
Anyone can propose an initiative by opening a GitHub issue in the CNCF TOC repo. The TAG will not activate an initiative without visible traction: at least three committed contributors is the informal bar. Initiatives target completion in roughly three months. Outputs can seed longer-lived sub-projects; the green reviews project, for example, will eventually become a permanent TAG sub-project. Contribution is mostly non-code work, publishing white papers and reference frameworks that the broader cloud-native community uses.
Q&A
How did observability end up under operational resilience rather than its own TAG? Tags are not exclusive; observability questions appear in every project graduation review under the day-two operations lens, which is why the TOC placed it here while cross-TAG collaboration remains open. ▶ 25:21
Does the three-month initiative limit cut off longer-term work? Initiatives kick off work and produce a defined output; successful ones spin out as sub-projects with no time limit, so the green reviews project will continue indefinitely after its founding initiative closes. ▶ 27:09
Is AI in operations covered by the TAG? AI topics are handled inside each category, for example measuring GPU energy consumption for projects like Kubeflow, rather than as a separate silo. ▶ 28:04
Notable Quotes
many of the CNCF projects which are in sandbox incubating or even a few graduated projects they wanted more guidance on how to formalize their release processes or structures on how do they manage code, how do they manage artifacts, how do they publish things, how do they distribute their artifacts, where do they distribute, how do they ensure supply chain security Nabarun Pal · ▶ 05:58
the more you utilize your infrastructure the lesser you will waste the resources and indirectly you are doing sustainability for your organization and in turn you are saving the costs as well Saiyam Pathak · ▶ 16:25
there were in the old text there were white papers that were started and were written for one and a half year and never published because they never got to the now we publish a state Mario Fahlandt · ▶ 27:48
i send a prompt and i there’s basically one tree less on on the planet Mario Fahlandt · ▶ 29:08
Key Takeaways
- The TAG requires at least three committed contributors before activating any new initiative.
- Green reviews aim to become a mandatory graduation requirement for all CNCF projects.
- Initiatives target roughly 90 days and can spin out permanent sub-projects on completion.
About the Speaker(s)
Nabarun Pal is a Principal Software Engineer at Broadcom, a Kubernetes project maintainer, and a former Kubernetes Steering Committee member. He chairs Kubernetes SIG Contributor Experience and has served as a Kubernetes Release Manager since the 1.21 cycle.
Saiyam Pathak is Principal Developer Advocate at vCluster and founder of Kubesimplify. He previously worked at Civo, Walmart Labs, Oracle, and HP across Kubernetes and machine learning infrastructure.
Alolita Sharma is an OpenTelemetry maintainer and Governing Committee member, a CNCF End-User TAB and GB member, and leads Apple’s AIML observability teams. She contributes to OpenTelemetry, Unicode, and W3C standards.
Carol Valencia is a Cloud Architect at KrolCloud with more than seven years in cloud-native technologies. She contributes to Kubernetes, OpenTelemetry, Kubernetes Release, and the CNCF Glossary.