Upgrading to a GxP Compliant Posit Stack with Kubernetes and Ansible for a Global Pharma Company

Project overview

‍

Appsilon partnered with a top 50 pharmaceutical company to upgrade their Posit stack so it would qualify for GxP compliance. The client needed the update to mitigate risks from a compliance perspective, avoid issues for data science teams, and make it audit-ready. Additionally, the client sought to leverage an internal Kubernetes cluster to scale platform resources.

However, their cluster version lacked full compatibility, requiring a customized approach to meet scaling demands. To avoid disruption of the workflows, Appsilon's team migrated the Posit environment instead of upgrading a legacy system. They used Ansible scripts to install and configure the new setup within a mirrored architecture, leveraging updated machines. After integrating the system with Kubernetes, they migrated configurations, applications, and user data. The upgraded environment is GxP compliant, audit-ready, and equipped with an improved change-management process.

‍

‍

Challenge

‍

When new internal GxP guidelines and requirements came into effect, the client's previously built Posit stack became outdated, posing a big risk from a compliance perspective. They needed to upgrade the system to the latest version and make it compliant as soon as possible. On top of that, the platform team on the client's side wanted to connect it to Kubernetes, but the older versions did not support it fully.

To solve the issue, the company searched for a trusted Posit partner and decided to go with Appsilon. When our architects scoped the project, we discovered additional bottlenecks:

• A simple upgrade of the Posit version to the latest one would disrupt the work of data teams and also would pose a risk of technical debt.
• Three working environments in the system were significantly modified and differed from one another. It posed a big risk for upgrades and migrations, requiring heavy customization on the installation and configuration.
• Data science teams’ limited understanding of software development best practices - such as dependency management, debugging, and version control - created a knowledge gap. This forced support teams to handle even routine technical requests, significantly increasing their workload.
• The GxP compliance setup required more work than initially expected

‍

‍

Solution

‍

Considering the client's requirements and limitations around them, Appsilon architects recommended migrating the Posit system with a set switchover date to make sure:

• The work of the data science team would not be disrupted.
• Three environments were merged to enhance consistency and reduce maintenance overhead. 
• The GxP process is followed from the ground-up.

Our team set up a new Posit system in a new environment, and then migrated apps onto it. We connected it to Kubernetes, making sure it's future-proof and scalable. We used Ansible as the tool for installation and configuration of the new environment and delivered the Ansible scripts on the client's code versioning tool. This allowed us to facilitate and improve the traceability, auditability, documentation and validation of the solution.

‍

Additionally, we supported the creation of qualification documentation and provided comprehensive technical input for the validation process. Finally, we delivered knowledge transfer sessions on Ansible, Docker, and Kubernetes, empowering teams to operate independently and significantly reducing the support team’s workload.

‍

‍

Impact

‍

The upgraded system meets all GxP compliance requirements, significantly reducing regulatory risks for the client. Choosing Ansible for compliance simplified the qualification process, and by implementing Kubernetes integration, we created a future-proof, scalable environment that can adapt to growing demands.

‍

Data scientists gained access to updated tools with zero workflow disruptions during the transition, requiring only minor adjustments for migrating a few apps due to outdated dependencies. The standardized infrastructure across all three environments established consistency and provided better control for the client's operations  team.

‍

Support operations improved considerably through our implementation of automation scripts for common tasks like package installations. This enabled faster changes and reduced the support team's workload for routine requests.

‍

Though the initial request was for an upgrade, our migration approach proved essential to maintain data science operations. The new environment successfully balances regulatory compliance with operational efficiency, creating a solid foundation for the client's data science initiatives moving forward.

‍