IBM Cloud Deployable Architectures for AI platform engineering

AI integration is vital for businesses to survive and grow in today's tech-driven landscape. IBM has integrated AI into many aspects of their daily operations, including customer support, HR and sales. It has enhanced efficiency by automating many tasks and optimizing processes. Previously, the CIO Hybrid Cloud Platform had faced a critical backlog of production AI applications.

The platform team endeavored to improve AI application productivity by creating application environments that follow best practices using in-house scripts and UI components. However, this complexity also introduced a few hurdles, for example, maintaining pace with rapid technological advancements while focusing on security, scalability and compliance. The dynamic nature of these applications necessitated frequent updates and modifications, straining engineering resources.

This situation presented a unique opportunity for IBM to devise a cutting-edge solution that could help bridge the gap between ambitious application requirements and resource limitations. IBM could optimize the use of their own technologies to revolutionize application management and deployment processes.

Deployable Architectures (DA) on IBM® Cloud® proved to be a great solution for this problem. A deployable architecture is a cloud automation solution that simplifies the process of setting up consistent, scalable and secure cloud infrastructures. By utilizing reusable modules—stand-alone units of automation code designed for specific infrastructure tasks—and combining them, teams can form a comprehensive deployable architecture, much like a pre-configured, automated blueprint.

The CIO Hybrid Cloud Platform team created custom DAs based on Terraform to create and maintain application environments. Terraform allows for standardization of infrastructure automation while managing the full infrastructure lifecycle across deployment, maintenance and decommissioning. With this approach, the platform team can employ the best infrastructure and workflow practices for application deployment so that developers can easily self-service at scale. The resulting DAs were added to the CIO Hybrid Cloud Platform enterprise account catalog, which enabled the application team to find, configure and deploy DAs in their sub-accounts. Over time, the platform team enhanced the DAs to disseminate new capabilities to the existing applications.

For example, one DA based on the IBM-supported retrieval augmented generation (RAG) DA module contained a standard application environment that could be deployed on the Red Hat® OpenShift® platform on IBM Cloud® and included various IBM-provisioned services, such as IBM watsonx.data®, IBM watsonx.ai®, Cloud Key Management, Cloud Secrets Manager, Cloud Object Storage, Cloud Logs, Cloud Observability Solution and others. A continuous integration (CI) toolchain was also created with IBM Cloud Continuous Delivery, encapsulating IBM’s best practices for AI application software development.

Deployable Architectures on IBM Cloud significantly boosted AI application onboarding. The IBM CIO observed five teams successfully onboarded with a base DA within the initial four weeks.¹ Most applications were based directly on a provided DA. Advanced application teams skilled at adapting their application environments were able to understand and modify a base DA. Constructive feedback from teams also drove immense improvements, with the existing DAs updated to newer versions—enabling applications to benefit from new features and bug fixes. This approach not only simplifies deployment but also aims to improve compliance postures to help ensure successful deployments across the board.

In conclusion, the IBM CIO reduced their average deployment time from 21 days to fewer than 11 days within the first month (August) by reusing a base DA.² All in all, the IBM CIO achieved 7x faster application deployment using a base DA compared to the previous non-DA deployments.³ This improved the way applications were deployed. As a result, the IBM CIO deployed 12 applications in six months using a base DA.⁴

The IBM Chief Information Officer (CIO) organization leads internal IT strategy and delivers, secures, modernizes and supports the IT solutions that employees, clients and partners use to do their jobs every day.