For years open source software has been deployed on z/OS using the System Modification Program/Extended tool, or SMP/E, to align with the package management system of z/OS.  With the release of z/OS Container Extensions (zCX), we provided open source content in the container format, and access to it through common command line interfaces like Docker and Podman.  When the Python AI Toolkit for IBM z/OS shipped in early 2023, the contents of the product were delivered as Python packages through the Python package manager (pip).  In both of these cases, we used native open source infrastructure to deliver and install the product rather than wrapper the content in z/OS-specific packaging and delivering through SMP/E.

This transition to open deployment mechanisms on z/OS is essential to adoption of new workloads on the platform.  In many cases these workloads are more than just applications built for a given operating system.  Open source developers see their platform as operating system + infrastructure.  When we deliver packages without their associated command line interfaces and other management infrastructure, users don't know how to proceed.  This makes z/OS different, and no one has time to sort out the necessary changes to make these workloads function properly.

Those with z/OS backgrounds will understandably have a lot of questions about things like security, and how to properly keep control of software as it is updated over time.  These are important questions, and we believe there are solutions in the open source community to address them.  We have recently taken a deep-dive into secure open source deployments for the Python AI Toolkit to describe the important details of an enterprise environment that leverages open software management infrastructure, while meeting typical z/OS qualities of service.  Please see

https://github.com/ibm-z-oss-oda/python_ai_toolkit_zos/blob/main/docs/README.md

for more details, and feel free to comment or ask questions in our Toolkit discussion group.