Modernizing a Windows-Based ERP: Elastic Beanstalk to Amazon EKS
A customer-facing Windows ERP moved from Elastic Beanstalk to Amazon EKS in Windows containers, without a rewrite: predictable auto-scaling, zero-downtime releases and full-stack observability.
- Modernization
- Manufacturing & Distribution
- AWS
- rolling releases
- Zero-downtime
- auto-scaling at peak business hours
- Predictable
- observability with APM, Prometheus and Grafana
- Full-stack
At a glance
- Engagement
- Modernization
- Industry
- Manufacturing & Distribution
- Platform
- AWS
- Region
- Not published
- Client
- Enterprise manufacturing and distribution ERP operator, name withheld
- Technologies
- 11 in the delivered stack
Before and after
A business-critical, customer-facing Windows ERP on Elastic Beanstalk with RDS for SQL Server slowed down during peak hours. Performance problems reached customers directly, and the team had no application-level visibility into why.
Before
- Elastic Beanstalk gave limited control over auto-scaling, and load balancing distributed traffic unevenly.
- There was no APM and little observability, so degradation had no clear root cause.
- The ERP had to be modernized without rewriting the application.
After
- Zero-downtime rolling releases
- Predictable auto-scaling at peak business hours
- Full-stack observability with APM, Prometheus and Grafana
The solution
Solution path
- Elastic Beanstalk
- Windows containers
- Amazon EKS + HPA
- ALB + Ingress
- RDS for SQL Server
How it was delivered
Phase 1
Containerized the ERP with Windows containers and moved it from Elastic Beanstalk to a production-grade Amazon EKS cluster with HPA, an Application Load Balancer and Kubernetes Ingress.
Phase 2
Kept RDS for SQL Server, tuned it, and replaced static credentials with IAM Roles for Service Accounts and least-privilege access.
Phase 3
Added APM, Prometheus, CloudWatch and Grafana with centralized logging and alerts; provisioned everything with Terraform and enabled rolling, zero-downtime releases.
Technology stack
- Amazon EKS
- Windows Containers
- Docker
- Amazon RDS for SQL Server
- Application Load Balancer
- Kubernetes Ingress
- IRSA
- Prometheus
- Grafana
- Amazon CloudWatch
- Terraform
Questions about this project
Did the ERP application have to be rewritten?
No. The Windows-based ERP was containerized with Windows containers and moved to Amazon EKS as it was, so the business gained scalability and observability without a rewrite.
How was the peak-hour slowdown fixed?
Kubernetes horizontal pod autoscaling gave precise control over scaling, an Application Load Balancer with Kubernetes Ingress fixed uneven traffic distribution, and APM made slow transactions visible so they could be addressed.
Related services
Related technical articles
Planning something similar?
Whether you're migrating, cutting costs, securing your infrastructure or exploring startup programs, start with the Cloud Readiness Survey.
What happens next
- Answer six questions about your cloud setup, spend and priorities. It takes about three minutes.
- A Celestibia engineer reviews your answers within one business day.
- You get tailored recommendations and the AWS programs you are likely to qualify for.
- sales@celestibia.com
- Phone
- +91 8308535810
Monday to Friday, 10 AM to 7 PM IST