Context
The group runs retail stores, a corporate office and a warehouse on a central Microsoft ERP hosted in its own data center: rack servers, a SAN and a Microsoft Hyper-V virtualization platform. Every store transaction, purchase order and report depends on that environment being available.
Business challenge
- The host platform and virtualization layer needed an upgrade for performance, security and application compatibility.
- ERP application servers had to stay available through maintenance and hardware failures.
- Updates, backup and monitoring were largely manual or reactive.
- The data center itself had to move to a new physical site without disrupting operations.
Architectural approach
- 01
Upgrade the platform underneath the ERP
Upgraded the main host’s operating system and Hyper-V environment, and optimized virtual machines for the applications, databases and services they run.
- 02
Cluster the business-critical servers
Managed and optimized Windows Server Failover Clustering for the ERP virtual servers, so a single host failure does not take the ERP down.
- 03
Centralize patching and supervision
Deployed WSUS for centralized operating system and application updates, and a PRTG monitoring server for proactive alerts on the infrastructure.
- 04
Add a dedicated backup tier
Deployed new NAS-based backup infrastructure for the server environment.
- 05
Extend secure remote access
Published the ERP through RemoteApp and configured Remote Desktop access for external teams, with user profiles provisioned for domain and ERP access.
- 06
Relocate the data center
Coordinated and executed the move of the data center to a new physical site.
Results
- ERP virtual servers run on a clustered platform, improving resilience and business continuity.
- Operating system and application updates are administered centrally instead of machine by machine.
- Infrastructure issues raise proactive alerts rather than being discovered by users.
- The data center operates from its new site.
- Remote and outsourced staff reach the ERP through published applications instead of full network access.
Lessons learned
Availability is built in layers: clustering protects the ERP, but only together with patching, backup and monitoring does it become an operating model.
A relocation is a business continuity exercise first and a logistics exercise second.
Platform versions reflect the environment at the time of the project. New designs start from currently supported platforms.