The context
The City of Geneva operates several dozen business applications serving more than 200,000 residents: citizen portals, internal management applications, communication services, payment platforms for public services. Like every Swiss public administration, it has to reconcile three imperatives that often pull in opposite directions: continuous infrastructure modernisation to deliver quality digital services, absolute data sovereignty under Swiss jurisdiction and ideally inside a controlled operational perimeter, and the autonomy of its in-house IT teams who cannot remain durably dependent on an external provider for daily operations. The challenge presented to Hidora in 2024 was precisely that equation: how to modernise without outsourcing, how to industrialise without losing control, how to accelerate without compromising compliance.
The technical challenge
Before the Hidora engagement, the City's application infrastructure was a patchwork of traditional VMs, hand-written deployment scripts from different eras, and a lightweight orchestrator showing its limits against the growing application portfolio. Three concrete operational issues stacked up. The average time to deploy a new application version reached 4 to 6 hours, with a share of manual handling that made the procedure fragile and stressful for the teams. Consistency across environments (development, pre-production, production) relied on operator institutional memory rather than versioned configuration, regularly causing post-deployment incidents. Finally, global observability across the application fleet was limited to basic per-application probes, with no consolidated view of overall behaviour.
The Hidora solution
The target architecture is built on a managed Kubernetes cluster deployed on Hikube, the Swiss sovereign cloud operated by Hidora in Geneva. That choice guarantees by construction that all data stays on Swiss territory, under Swiss jurisdiction and operations, and that the underlying infrastructure is itself controlled by a Geneva-based entity. On top of Kubernetes, we rolled out a standardised CI/CD chain based on GitLab CI, with reusable pipelines across applications, YAML manifests versioned in Git as the single source of truth, and a complete observability stack (Prometheus + Grafana + Loki) accessible to all of the City's application teams. The migration work spread over 6 months in progressive mode: a first non-critical application served as proof of concept and training ground, followed by a controlled scaling phase during which the internal teams took ownership of the platform with Hidora support.
The measured results
Six months after kick-off, operational indicators confirm the impact. The average deployment time for a new application version dropped from 4 to 6 hours down to roughly 30 to 50 minutes depending on complexity, a reduction in the order of 70%. The overall availability of public-facing services sits at 99.9% over the last 12 months, with a clear drop in post-deployment incidents thanks to guaranteed cross-environment consistency. Data sovereignty is demonstrable during audits: 100% of workloads run on Swiss servers operated by a Swiss team, with full traceability of accesses and changes. And crucially, the City's IT team now operates the platform itself daily: Hidora remains available for advanced architectural questions and strategic evolution, but the operational run has been handed over as planned at the end of the engagement.
The transferable lesson
This project illustrates a trajectory we regularly observe in the Swiss public sector: modernisation does not require deep outsourcing or an American public cloud to succeed. With a sovereign Kubernetes platform, standardised DevOps practices and a carefully designed knowledge transfer, a mid-sized administration can reach a level of operational maturity equivalent to that of large tech companies while strictly meeting its sovereignty and autonomy obligations. The key factor is not the technology itself (Kubernetes is widely adopted), it is the way it is deployed: progressively, with continuous training of internal teams, and with an operator who understands the specific constraints of public administration.