Overview
ReUpMobile is a wireless dealer management platform built on .NET and ASP.NET, serving wireless retailers across the United States. Their legacy deployment process included manual IIS configuration, RDP-based server management, and multi-hour release windows, which was unsustainable as the business scaled. Every deployment was a risk, scaling required days of infrastructure provisioning, and the team was told by multiple consultancies that the only path forward was a complete rewrite to Linux. Dcode proved otherwise.
The Challenge
Running Windows workloads on Kubernetes is not a standard playbook. Windows containers have different networking models, multi-gigabyte base images, and limited community tooling compared to Linux. Most consultancies either refuse Windows container projects or insist on a full rewrite. ReUpMobile had years of battle-tested .NET business logic. A rewrite would take months and introduce regressions. They needed modernization without starting over.
Deployments took 2 hours of manual work via RDP sessions. Provisioning a new environment took 3 days. Scaling required manual server configuration and load balancer updates. The team had no CI/CD pipeline, no infrastructure as code, and no automated scaling, so every operational task was a manual process.
Deployments: 2 hours manual via RDP. New environments: 3 days to provision. Scaling: fully manual. Every release was a production risk with no rollback capability. The team was spending more time on operations than development.
Our Solution
Dcode built a multi-zone Amazon EKS cluster with dedicated Windows node groups running across multiple availability zones for high availability. The .NET and ASP.NET applications were containerized using Windows containers with the same code and same binaries, no rewrites. They are now orchestrated by Kubernetes with health checks, rolling updates, and automatic recovery. Karpenter manages Windows node provisioning, scaling the cluster based on actual demand.
We implemented Horizontal Pod Autoscaler (HPA) with custom metrics to automatically scale application pods based on request rates and response times. Helm Charts standardized deployments across environments, and a full CI/CD pipeline was built so the team could ship with confidence. The entire infrastructure was codified with Terraform. What used to take 3 days to provision now happens in minutes with a single command.
Amazon EKS supports Windows node groups alongside Linux nodes in the same cluster. This hybrid approach lets Windows applications run containerized with Kubernetes orchestration while Linux nodes handle cluster services, monitoring, and ingress controllers. No application rewrite required; existing .NET binaries run as-is in Windows containers.
Results
- 92% deployment time reduction, from 2 hours of manual RDP work to 15-minute automated Helm releases
- 95% reduction in manual scaling operations through HPA and Karpenter handling capacity automatically
- 99% infrastructure provisioning reduction, with new environments in minutes via Terraform instead of 3 days manually
- Zero application rewrites: existing .NET/ASP.NET code containerized and running on EKS Windows nodes
- Multi-zone high availability with automatic failover and self-healing via Kubernetes orchestration