Kyverno: The Reference Architecture for Kubernetes-native Policy Management
TL;DR Kubernetes is permissive by default: it allows developers almost anything, including insecure …

The goal of the Online Access Act is ambitious: all administrative services should be digitally available. However, the implementation often failed in the first phase due to the fragmented federal structure. The solution for the second phase (OZG 2.0) is the EfA principle. One state develops a service (e.g., the parental allowance procedure) and provides it centrally as a cloud service for all other municipalities.
Technically, this is a massive scaling and integration problem. It requires a shift from local “island servers” to a modern, multi-tenant cloud-native platform.
For a service like “vehicle registration” to work simultaneously for thousands of municipalities, the infrastructure must be based on three technical pillars:
In an EfA architecture, many municipalities share the same physical infrastructure, but their data must remain strictly separated.
The core problem of administration is interoperability. An online application must be securely transferred to the respective specialized procedure of the municipality.
When a nationwide relevant service (e.g., an energy price allowance) is launched, the infrastructure must be able to scale from zero to millions of users within seconds.
In the public sector, every change must be documented. Here, the GitOps approach (e.g., with ArgoCD or Flux) offers an invaluable advantage: the entire configuration of the administrative infrastructure is versioned in a Git repository.
What is the difference between a “normal” cloud and an OZG cloud? An OZG cloud must meet the requirements of the BSI basic protection and the confidentiality protection order in addition to the technical standards. Moreover, the connection to the “administrative network” (NdB - Networks of the Federal Government) is a specific technical requirement to securely reach internal specialized procedures.
Why are Containers (Docker/Kubernetes) used for EfA services? Containers ensure that software runs exactly the same in the development environment as it does later in the service provider’s data center. This prevents the “It worked on my machine” problem when handing over software between different authorities or IT service providers.
How is data protection technically guaranteed for EfA services? Through a combination of technical isolation (Namespaces, Micro-Segmentation) and organizational separation. Additionally, technologies like “Confidential Computing” are often used, where data remains encrypted even during processing in memory.
How do municipalities connect their local specialized procedures to the EfA platform? Usually via secure VPN tunnels or dedicated lines to the administrative network. Communication takes place over standardized APIs (like XÖV standards), ensuring that the data formats between the central online service and the local database are compatible.
Can EfA services also be hosted with US cloud providers? This is legally controversial (keyword: Schrems II). The technical strategy therefore strongly leans towards sovereign platforms under German jurisdiction or private clouds of public IT service providers (like Dataport, Komm.ONE, or IT.NRW) to maintain full sovereignty over citizen data.
TL;DR Kubernetes is permissive by default: it allows developers almost anything, including insecure …
TL;DR Secrets (API keys, database passwords) do not belong in Git code, but their runtime …
Until now, monitoring was often a compromise: Those who wanted to know exactly what was happening …