Kubernetes & Infrastructure
Efficient deployment of ERP software to 100 opticians
To replace an inefficient infrastructure, we helped the GMG group build a robust, automated and scalable Kubernetes solution for rolling out their ERP software quickly to 100 opticians.
The project: efficient deployment of ERP software to 100 opticians
The GMG group needed a robust, automated infrastructure to roll out its ERP software for opticians to 100 member businesses. The existing system ran on Windows servers, was time-consuming and therefore inefficient. The goal was a solution that minimises effort, scales, and supports the opticians’ business processes reliably.
The solution: DevOps expertise from devs group
Kubernetes and on-premise infrastructure: we laid the foundation for the new solution by setting up the Kubernetes cluster entirely through Ansible and a pipeline. New nodes are added through a configuration entry. GMG can scale the cluster up and down themselves without significant effort.
Infrastructure as code: the build of the cluster is documented and versioned in Ansible. That makes the configuration repeatable and produces an error-free environment, which translates into a clear gain in efficiency and a saving in resources.
CI/CD: by adopting GitOps practices along with ArgoCD and Helm, we automated deployments and made them visible. That gives a clear overview of which software version is installed at which optician, and produces fast, dependable rollouts.
Observability: to monitor system performance we implemented a comprehensive observability stack with Grafana, Prometheus (for metrics) and Loki (for logs). Potential problems surface early.
The result: quality and efficiency that speak for themselves
Working with devs group let the GMG group solve the problem of laborious deployment and simplify its processes.
Frequently asked about this project
Why on-premise rather than cloud?
Because the opticians’ ERP software should run where the group’s data and existing system landscape are. Kubernetes on-premise combines both: control over your own infrastructure with the automation usually associated with the cloud.
How does a new node join the cluster?
Through one line of configuration. The whole cluster is described in Ansible as infrastructure as code and set up through a pipeline; GMG scales up and down on their own, without having to call us. That was precisely the goal.
What did GitOps actually buy here?
Visibility and repeatability. With ArgoCD and Helm it is visible for each of the 100 member businesses which software version is deployed; a rollout is a Git commit rather than a maintenance window on Windows servers.
