Jun 13
KubeCon Europe 2022 — our highlights
This year, ADEO gave us the chance to participate in the biggest Kubernetes event worldwide. This allowed us to better understand the variety and the depth of the community around CNCF projects!
With more than 7000 participants, KubeCon Europe is huge! 65% of the participants were at their first KubeCon.
Resolutely focused on the values of open source and the community, the CNCF is constantly looking for partners, developers and contributors in the broad sense. A contribution can be simply reporting a bug on Github!
For me, this KubeCon seems to be oriented on security. With the growing threats and the increasing complexity of technology and architecture, everyone must now verify, scan and protect themselves…
Today the CNCF proposes more and more standardized solutions via a large number of interfaces. KubeCon was an excellent opportunity for us to see them.
❤️Here are our favorite sessions ❤️:
- Effective Disaster Recovery: The Day We Deleted Production — Rick Spencer & Wojciech Kocjan, InfluxData
- Network-aware Scheduling in Kubernetes — José Santos, Ghent University
- Seeing is Believing: Debugging with Ephemeral Containers — Aaron Alpar, Kasten
- From Kubernetes to PaaS to … Err, What’s Next? — Daniel Bryant, Ambassador Labs
❤️Effective Disaster Recovery: The Day We Deleted Production — Rick Spencer & Wojciech Kocjan, InfluxData
This is a feedback on a major incident at InfluxData where ArgoCD and the GitOps approach deleted all production data resulting from bad configuration…
It should be understood that influxData, as a database host, has all of its infrastructure in Kubernetes mode, including the volumes (PV).
The crux of the presentation is the incident origin. Rick and Wojciech explained to us that they have merged a Github PR that only adds Kubernetes resources and that ArgoCD will “simply” create them. But in reality, this action deleted all the data. Emerging from such a horror story, everybody was wondering if their own operations are really 100% ok: git flow, release management, staging environment, review etc… We also find the timeline hour by hour and the major decisions that were taken; including the first decision “we don’t touch anything until we have a real action plan!”. To see or see again without hesitation.
❤️Network-aware Scheduling in kubernetes #OPS
The native Kubernetes scheduler allows us to reduce infrastructure costs by using resources more efficiently. For applications with an end-to-end latency which must be as low as possible, the scheduler which is limited to CPU/Memory resources, does not take the bandwidth into consideration (especially on regional networks). In such cases, it is possible to deploy a specific scheduler that takes into account network performance and topologies.
Kubernetes SIGs: scheduler plugins based on the scheduler framework
❤️ Seeing is Believing: Debugging with Ephemeral Containers — Aaron Alpar, Kasten
Very interesting talk on the best way to debug a pod without being intrusive and impacting, based on linux namespaces. This feature will be delivered in Kubernetes v 1.23, remind yourself once it’s available !
❤️From Kubernetes to PaaS to … Err, What’s Next? — Daniel Bryant, Ambassador Labs
This talk focuses on the complexity of the K8s ecosystem for a developer and the need to position themselves as a platform to facilitate the developer experience… great similarities with what we experience at Adeo. For example, the platform must be considered as a product in its own right (Golden Path).
Check the full list of talks here.
See you next year round ! (KubeCon 2023 17–21 APRIL AMSTERDAM, THE NETHERLANDS)
Christophe, Franck, Maxime, Xavier.
