Frankfurt runs cloud without a Kubernetes team

Frankfurt’s Public Health Department runs GA-Lotse, a federated platform for public health applications. The standout feature? The authority uses a sovereign, encrypted platform without needing to build its own Kubernetes specialist team.

Key Takeaways

Related:BSI C3A: Cloud sovereignty becomes verifiable  /  Europe’s first exascale supercomputer: who gets access?

Expertise shouldn’t depend on the cluster

Public health offices need digital systems that can handle day-to-day tasks like examinations, scheduling, and inspections. But they don’t need to train their own Kubernetes platform teams to run these systems securely. GA-Lotse separates these two responsibilities.

Frankfurt’s Public Health Department modernized its infrastructure using Red Hat OpenShift as a managed service. The platform is federated, featuring access control and encryption. Red Hat published the case study on July 6, 2026.

Three months to go live

According to the operating partner VSHN, the contract was awarded in August 2024, and the platform went live three months later. This timeline doesn’t replace project analysis, but it shows what happens when hosting, platform operations, and application development aren’t all dumped onto a single overstretched team.

GA-Lotse is now in production across multiple regions in Hesse, supporting applications for school entry examinations and dental screenings, among others.

Managed operations create breathing room

The key takeaway is about responsibility. A managed service doesn’t relieve an authority of architectural decisions or data protection obligations. But it does consolidate the specialized knowledge needed for platform operations, updates, and daily maintenance. This frees up subject-matter teams to focus on the processes they’re actually responsible for.

For many public-sector teams, this approach is more realistic than building an in-house Kubernetes setup. The question isn’t whether containers are technically manageable-it’s which expertise needs to stay on-site long-term and which makes more sense as a clearly defined operational service.

Impact with transparent sourcing

VSHN reports that GA-Lotse supports 7,000 school entry examinations and 45,000 dental screenings annually. These figures describe the processes being digitized-not independently measured productivity metrics. This distinction matters: the case study provides a credible operational narrative, not a one-size-fits-all benchmark for every authority.

The platform replaces paper-based steps in the mentioned applications. However, the available sources don’t allow for a reliable assessment of time savings at individual offices.

Open Source Code Prevents the Next Reboot

GA-Lotse has been published on openCode.de. This allows the solution to be reviewed and reused beyond Frankfurt. Open source code doesn’t solve procurement or operational questions on its own-but it does change the starting point. Other public health authorities no longer have to start from scratch with every platform idea.

The case remains a vendor and partner case study, not an independent benchmark. Yet as a Reboot Germany story, it still delivers by making the operational split visible: expertise, platform operations, and reuse interlock. The Red Hat case study outlines the technical framework.

For other authorities, this creates a pragmatic review sequence. First: Which professional data and roles need to be mapped on the platform? Second: Which operational services can be reliably outsourced to a specialized partner? Third: Which parts of the code and interfaces should remain reusable beyond the initial project? Answering these questions before procurement reduces long-term dependencies.

The advantage of a managed service isn’t blanket relief-it’s transparent accountability. The professional side retains process knowledge and priorities, while platform operations handle recurring technical tasks. Both sides need clear handovers and a robust escalation process when an application falters under load or deviates from expected performance during an update.

Frequently Asked Questions

Image source: AI-generated (July 2026)