Referenz

VYDA GmbH

Information technology

Person maintaining servers

Long-standing client

VYDA, formerly Eucon Digital, is a provider of data-driven process automation for the insurance and property sectors. We have been advising and supporting VYDA for many years on architecture, IT operations and the further development of its IT infrastructure.

GitOps with FluxCD

A key component of our collaboration was the introduction of FluxCD as a GitOps tool. The aim was to deliver service releases and deployments in the Kubernetes cluster in an automated and reproducible manner. In addition, we migrated Jenkins to GitLab CI to standardise the release process within a unified environment.

5
Project duration in months
100+
Repositories adapted
7
Clusters centrally managed
100%
GitOps & Compliance-driven

Project

Introduction of FluxCD as a GitOps tool

The following account focuses on the introduction of FluxCD. We were responsible for the design, planning and implementation of the entire project. The individual steps were coordinated and prepared in collaboration with the VYDA team to ensure that operations could continue smoothly during the roll-out.

Starting point

Prior to the introduction of FluxCD, VYDA used a Kubernetes cluster with applications deployed as Helm charts. Deployments were carried out via a Jenkins CI pipeline, which applied changes directly to the cluster. VYDA managed the Helm charts via a Git repository; however, manual changes to the cluster went unnoticed. This posed a risk of configuration drift. Should the cluster become corrupted, recovery was only possible to a limited extent.

In addition, we introduced SOPS to manage sensitive data such as API keys and passwords. This ensures that such data is stored in the Git repository in an encrypted form, in accordance with GitOps standards.

Challenges

The introduction of FluxCD required the migration of the existing configuration to a new structure. The aim was to be able to manage several Kubernetes clusters in parallel via a shared Git repository. Furthermore, the existing deployment process was to be switched from a ‘push’ approach via the CI pipeline to a ‘pull’ approach using FluxCD. To achieve this, the CI pipeline had to be adapted across several hundred repositories. To avoid errors and ensure business continuity, we implemented these changes automatically.

In parallel with the introduction of FluxCD, the continuous integration system was migrated from Jenkins to GitLab CI. During the migration, operations had to continue running stably and without interruption.

Milestones

  1. Development of a configuration structure that supports multiple Kubernetes clusters with numerous Helm charts and other Kubernetes configurations in a shared Git repository.
  2. Development of centralised GitLab CI pipeline templates for container building and for updating container tags in the cluster configuration’s Git repository.
  3. Migration of all Jenkins pipelines to GitLab CI.
  4. Shutdown of Jenkins.
  5. Rollout of FluxCD in the existing cluster and migration of the existing Kubernetes configurations to FluxCD.

Technologies

We have chosen FluxCD as our GitOps tool because it offers a wide range of features whilst requiring minimal resources:

  • Native support for Helm Charts.
  • The ‘pull’ principle reduces the attack surface of the CI pipeline.
  • Reliable synchronisation mechanism.
  • Automatic deletion of resources that are no longer required.
  • Integration of SOPS for encrypting sensitive configuration data.

All the technologies used are licensed under open-source licences. This keeps operating costs low whilst also allowing errors to be traced down to the finest detail. In the case of FluxCD, we reported issues directly to the project and resolved them in collaboration with the community.

The result

Stable in operation, automated and reproducible

We developed extensive automation processes for the adaptation of several hundred repositories. This enabled us to prepare the necessary changes, verify the results and resolve edge cases manually in a targeted manner.

The migration to FluxCD went smoothly. The new configuration structure has proven its worth and now manages seven clusters with around 100 applications, operating stably.

Configurations are shared across clusters. This significantly reduces the effort required for maintenance and patch management.

Following the FluxCD migration, one cluster was moved to a new cloud infrastructure. This went smoothly, as the entire cluster configuration could be imported from the Git repository.

The entire cluster configuration is automatically validated with every adjustment. In addition, we have integrated further Kubernetes compliance tools such as Kubescape.

Clear documentation and standardised processes enable work to be carried out on the cluster systems in parallel. All configurations are maintained in the Git repository, ensuring they remain traceable and reproducible. Changes are prepared via pull requests and checked before rollout.

" Working with identinet GmbH has transformed our infrastructure and development processes. From implementing GitOps using FluxCD to migrating from Jenkins to GitLab CI, they have modernised our entire DevOps pipeline. Their expertise in automating complex microservice deployments and in setting up our OpenTelemetry-based monitoring infrastructure has been invaluable. Beyond the technical implementation, their consultancy has driven innovation in our operational processes. A trustworthy and reliable partner for critical infrastructure. "


VYDA GmbH

Max Reichert

Head of Software Development @VYDA GmbH

Summary

Sustainability and empowerment

The objective of improving service releases and deployments has been achieved. The seven environments – a shared infrastructure cluster, as well as development, staging and production clusters in two business units – are now operated using FluxCD and GitLab CI.

As our collaboration with VYDA progressed, projects emerged that, for example, strengthened information security, automate configuration management and identify issues in the operational infrastructure. Among other things, a centralised backup system and a cluster to centralise the operation of infrastructure services were introduced. Furthermore, we designed and implemented a central monitoring platform based on the Grafana stack.

Today

Together with the VYDA team, we continue to work on the further development and maintenance of the infrastructure. Our documentation supports both the teams managing the clusters and the developers deploying their applications via Helm Charts.

Schedule an initial consultation

Future-proofing IT infrastructure and IT services