Platform Engineering at CNS Munich

Xavier Rakotomamonjy — Tue, 26 Aug 2025 14:33:58 +0000

This year Cloud Native Summit (CNS) in Munich gathered adopters and technologists from open source and cloud native communities.

Compared to last year's Kubernetes Communities Days (KCD), the agenda dedicated many slots to Platform Engineering. This discipline can be seen as a way to fill gaps between different stakeholders of an Internal Developer Platform (IDP). Last year I wrote an article on this topic at the KCD Munich. This year I would like to share fresh ideas or principles from some talks given during the conference.

Identifying problems before solutions

Day 1 opened with a presentation of "Product thinking" approach made by Stéphane Di Cesare (Deutsche Kreditbank) and Dominik Schmidle (Giant Swarm). They reminded some core principles and techniques of "Platform Engineering" to manage the link between problem space and solution space.

Day 1 also closed with a panel discussion on "Platform Engineering" animated by Max Körbächer (Liquid Reply), focused on the adoption of IDPs by end users and within projects.

Video 1 Platform Thinking

Video 2 Panel Discussion

AI workloads on K8s

"Use platform engineering for AI workloads on K8s", that was the motto of the presentation given by Mario-Leander Reimer (QAware GmbH). He tackled emerging challenges for deploying Agentic AI workloads and introduced an example of what a platform should cover including quality plane and compliance plane. His talk is part of an open source initiative launched at QAware this year to mature the concepts of a dedicated control plane: the agentic layer.

Video 3 Architecting and Building a K8s-based AI Platform

Composable platforms on Kubernetes

Hossein Salahi (Liquid Reply) presented a reference implementation to streamline the delivery of services for developers. Some of the challenges were to link dev and ops, to manage services in configuration, to orchestrate infrastructure deployment and to manage access. For that purpose the open source tools Kratix codify a contract between dev and ops. Combined with Backstage, this abstraction allows services to be packaged and delivered to developers.

Video 4 Modular Platform Engineering with Kratix and Backstage

Reduce cognitive workloads

Interestingly the problem of cognitive workloads due to some platform complexity was a recurring theme. But Michel Murabito (Mia Platform) positioned platform engineering as a way to reduce cognitive workloads. He highlighted essential building blocks and features that can reduce mental burden.

Video 5 Creating a smooth Developer Experience

Involved user as contributor

Lian Li (lianmakesthings) told us a story of what can go wrong when building a platform based on real examples and some facts. She advocated some receipts that can improve communication among teams in an organisation and explain some benefits of having contributing users in the loop.

Video 6 Many Cooks, One Platform: Balancing Ownership and Contribution for the Perfect Broth

Convergence

Evelyn Osman (enmacc) discussed the importance of convergence when building a platform in order to avoid fragmentation in teams and solutions. She shared some important steps that foster innovations and better meet user expectations.

Video 7 Convergence on Platforms

Conclusion

This is not an exhaustive list of what was discussed, but somehow it made it clear that Platform Engineering approach can help picking the right solution in the open source landscape. Some enablers already provide a strong integration with Kubernetes to develop and publish new services on top of it. And depending on the organization, some shift in practices also might be required to follow the evolution of the platform at build and runtime. The conference was a great place to exchange with the community on technical and organization challenges.

References

Platform engineering at KCD Munich

Xavier Rakotomamonjy — Fri, 09 Aug 2024 10:08:25 +0000

Introduction

During the Kubernetes Community Days (KCD) in Munich, Camptocamp hosted a booth during the conference. This event also provided opportunities to see some great presentations and workshops on security, monitoring, network, and CICD.

One particular focus was given to the Platform Engineering concept. The term might not be new, but it is gaining traction. Back in the days, when it was first coined, the need for a new term was probably questionable. And the overlap with some already existing definitions such as Software engineering or Site Reliability engineering could have brought some confusion.

Through several presentations given from different actors (industry, service provider and software vendor), the conference shed new light on the topic of Platform Engineering. Instead of trying to give a definition of what it is, let’s try to summarize some of the subjects that were addressed.

Keep track of the user needs

Stéphane Di Cesare presented ongoing works in DKB to give the control back to the user with a focus on values. As the first result, values of a service are grouped into sets called capabilities. This approach aims at capitalizing development effort (e.g.: documentation) and to reduce cognitive loads by managing complexity of the platform regarding security and compliance.

Identify industry patterns

Another presentation was performed by Daniel Bodky. He introduced the Cloud Native Operational Excellence (CNOE, pronounced "Kanoué") as a framework whose primary focus is to standardize infrastructure.

In order to achieve this goal, stakeholders are divided into different categories. And on the other end, the platform is also divided into building blocks called technology capabilities. Capabilities may not fit the needs of every user, and CNOE introduced these layers to help companies of different sizes to improve their decision making in the CNCF technological landscape and to improve the Devops experience with open source solutions.

As an example of a result, he showcases the tool idpBuilder. The utility can build a complete environment based on Argocd, Gitea, Backstage, Keycloak, CrossPlane, external Secrets operator. The stack can run on a local machine, and the demo illustrated a templating approach with Backstage to create a catalog of new infrastructure services for the end user.

Idpbuilder create

Function as a Service

Mauricio Salatino and Thomas Vitale highlighted during their presentation the separation of concerns. By combining technologies in the field of FaaS (Function as a Service), such as Knative and Dapr, they demonstrated an example of development based on serverless infrastructure. The goal is to faster onboarding of application developers and decrease the load in a distributed environment context. Dapr enforces best practices for development and manage ready to use function in the following area:

From project to product

David Sterz, Stefan Schweer and Tim Schrumpf shared how Vorwerk planned to consolidate more than 10 years of technical changes of their infrastructure by adopting a platform-focused approach. The platform initially started on-premise with a European scope and went progressively worldwide and on the cloud to reach out millions of end-users. The needs of the platform also grow with different concerns relating to security or finops. The introduction of a platform as a product and not as a project is a paradigm shift to better handle platform maturity and future evolution and needs.

Conclusion

This post was not intended to give an exhaustive list of what was discussed on Platform Engineering, but rather to highlight some chunks that are part of it: create value for the user, reusable capabilities, tools and methods, platform evolution and maturity. All these lead to a better integrations of works. And from that perspective, the platform engineering concept might be seen as an additional effort to improve the continuum in different product life cycle development phase.

Some references

Presentation titles during the conference

Platform Engineering: DevOps Evolution or a Fancy Rename? by Dotan Horovits
Introducing a product mindset in a container platform team by Stéphane Di Cesare
Towards Standardized Platforms: How the CNOE Project can Help? by Daniel Bodky
Saturating people not systems: Lessons learned from building a platform to serve dishes worldwide David Sterz / Stefan Schweer / Tim Schrumpf
Unlocking New Platform Experiences with Open Interfaces by Mauricio Salatino and Thomas Vitale

DEV Community: Xavier Rakotomamonjy

Platform Engineering at CNS Munich

Identifying problems before solutions

AI workloads on K8s

Composable platforms on Kubernetes

Reduce cognitive workloads

Involved user as contributor

Convergence

Conclusion

References

Platform engineering at KCD Munich

Introduction

Keep track of the user needs

Identify industry patterns

Function as a Service

From project to product

Conclusion

Some references

Presentation titles during the conference

Other references