The client is an Australian stock transfer company that offers corporate trust, stock transfer, and employee share plan services worldwide.
The Challenge
To expand in the US, the client acquired corporate trust services from another financial company in 2021. This required migrating about 30 applications across various tech stacks, such as Java, .NET, WebSphere, Angular, and NodeJS. More than a hundred repositories also needed to be moved from the acquired company’s on prem infrastructure to the client’s VMware private cloud. The acquired unit had no DevOps culture, and the development team was inexperienced in building DevOps pipelines. It also lacked a standard release management process. Applications needed to be containerized and deployed to OpenShift. The client wanted the team to focus more on development rather than DevOps pipeline tasks. However, the team lacked knowledge of Kubernetes and containerization, and proprietary libraries weren’t shared. Retaining CI / CD artifacts on the cloud was expensive and required a cleanup mechanism, so it was necessary to educate the client’s team on DevOps principles.
The Solution
The client turned to Persistent for its cloud migration and DevOps expertise in several areas:
- Source Code Repository Migration: The team identified more than 100 source code repositories for migration to Azure DevOps Git. During the discovery phase, we outlined the external dependencies and proprietary libraries. We adopted a phased migration approach with dependencies uploaded to Azure DevOps (ADO) artifacts and configuration references updated.
- Templatizing pipelines: Persistent developed pipeline templates for different tech stacks to save time and reduce effort, covering build and publishing stages. ADO Task groups were established for deployment in environments like Dev and OpenShift. Sample build pipelines and guidelines enabled the client’s teams to create their own pipelines by choosing templates and parameters. This simplified DevOps, allowing teams to concentrate on development.
- Branching Strategy: In collaboration with the client’s global DevOps team, we implemented a branching strategy based on GitLab Flow, defining separate strategies for feature branches, releases, and patches. Global policies in ADO enforced code review checks, linked work items, and restricted direct commits. The strategy was documented in Confluence and shared with the development team.
- DevOps as a Service: We built an automated DevOps delivery model with CI / CD pipelines on ADO to accelerate the delivery cycle. Snapshot pipelines deployed development branches to Dev and SIT, while release pipelines deployed main branches to UAT and PROD. Separate templates for VM-based and containerized applications were created, with custom base images and Docker files. Environment-specific Kubernetes manifests were generated using a customization framework and GitOps and deployed to OpenShift via RedHat ACM.
The Outcome
The client can now:
- Adopt a layered approach for infrastructure provisioning.
- Leverage a common network layer consisting of VPC, subnets, and other network services.
- Improve cost visibility and tracking using a tagging strategy.
- Deploy highly optimized and configurable services.
- Provision, tear down, and schedule environments on demand.
- Save up to 20%–25% for non-production environments.
