By Open source maintainers targeting mainframe have had no reliable way to test their code. Mainframe hardware is expensive, highly specialized, and in many cases completely air-gapped from the internet. The result has been a structural bottleneck: only a handful of specialists at companies like IBM, Broadcom, and Rocket Software could do porting work, leaving the broader ecosystem unable to grow.
In this interview on TFiR, John Mertic, Director of Program Management at the Linux Foundation, walks through how the Open Mainframe Project has opened formal infrastructure access for any qualifying open source project, covering both z/Linux and z/OS environments, and what the application process looks like from start to finish.
Guest: John Mertic, Director of Program Management at The Linux Foundation / Open Mainframe Project
Show: Let’s Talk
LinkedIn: John Mertic
Watch the discussion on YouTube
Here is what every open source maintainer and platform engineer targeting mainframe needs to know.
Technical Deep Dive
Q: Why has mainframe infrastructure been inaccessible to open source projects for so long?
John Mertic, Director of Program Management at the Linux Foundation, explains that the problem dates back to at least 2015, when the Open Mainframe Project launched, and likely further. Mainframe hardware is not only expensive but highly specialized, and many systems are completely air-gapped from the internet, making remote access impossible. Mertic notes that stopgap solutions existed over the years but none fully solved the problem of providing accessible infrastructure to any open source project targeting the s390x platform.
“This hardware is not only very expensive and specialized, but it’s also not easily accessible.”
John Mertic, Director of Program Management, The Linux Foundation
Q: What makes mainframe architecture different from x86 or ARM for open source developers?
Mertic explains that mainframe is a distinct architecture with underlying hardware features and functionalities that are unique and custom-tailored compared to x86 or ARM. On the Linux side, the experience is similar to other platforms, but z/OS is an entirely different environment in terms of how developers interact with it, work with it, and support it from an operating system perspective. This architectural gap has made it difficult for open source maintainers to build and test without access to real hardware.
“z/OS is an entirely different animal. It’s unique in many ways, from how you interact with it and work with it to how you support it from an operating system perspective.”
John Mertic, Director of Program Management, The Linux Foundation
Q: What bottleneck did the lack of mainframe test infrastructure create for open source maintainers?
Without accessible infrastructure, open source maintainers responding to demand from mainframe customers had no way to test their code and were left guessing. In practice, the porting work fell to specialists at IBM, Broadcom, and Rocket Software, who would upstream changes back to the community. Mertic acknowledges this was never an efficient method and created a concentration of expertise that prevented the broader ecosystem from scaling. The Linux Foundation’s own hosted projects were also affected, unable to provide mainframe infrastructure to mentees or project maintainers who requested it.
“It’s really made it difficult for the larger ecosystem to take off.”
John Mertic, Director of Program Management, The Linux Foundation
Q: How does an open source project apply for mainframe infrastructure through the Open Mainframe Project?
Projects apply through the Open Mainframe Project Technical Advisory Committee (TAC). The TAC reviews applications against core requirements: the work must be an open source project, not commercial work, designed to benefit the mainframe community as a whole and grow the mainframe ecosystem. If approved, the TAC passes the project to backend teams at IBM, who handle provisioning. Mertic emphasizes the process is straightforward and does not require prior mainframe expertise.
“Apply to the TAC. The TAC reviews your application, approves or rejects it, and helps you move on to the next step in the process.”
John Mertic, Director of Program Management, The Linux Foundation
Q: What does a standard mainframe infrastructure offering include for open source projects?
Mertic describes a standard out-of-the-box offering that includes a standard z/Linux image and a standard z/OS image, ready to use without requiring applicants to specify IPL counts, disk space, or other technical details. This lowers the barrier significantly for maintainers who have no prior mainframe experience. Projects with more specific hardware or configuration requirements can request flexible accommodations through the partnerships the Open Mainframe Project has built with its members.
“You don’t have to know things like how many IPLs you need, how much disk space you require, and so on. It’s a very standard offering that’s accessible and easy to use for anyone.”
John Mertic, Director of Program Management, The Linux Foundation
Q: Who contributed to making this mainframe infrastructure program possible?
Mertic credits Broadcom, Converged Technologies, IBM, and other Open Mainframe Project members for contributions that included making hardware available, contributing software, and providing resources and time. He describes the program as the culmination of years of discussions, stopgap efforts, and community coordination, and calls it a first of its kind, noting that no program of this type had existed before.
“This is truly the definition of a community effort.”
John Mertic, Director of Program Management, The Linux Foundation
Resources & Documentation
- Open Mainframe Project, Linux Foundation project providing governance, community, and now infrastructure access for open source mainframe development
***
👇 Click to Read Full Raw Transcript
Swapnil Bhartiya: Let’s talk about this announcement where you just made a new mainframe infrastructure, resources available to open software we talked about in past. Finally it’s happening. So talk a bit about how much time it took for it and what does it mean for developers who target mainframe as their of course production systems.
John Mertic: So the Open Mainframe project launched a little bit over 11 years ago. So at 2015 LinuxCon in Toronto, we’ve been having questions about how do you get access to hardware to do open source project testing since at least that point and probably going back farther from there. This hardware, not only is it just very expensive, it’s very specialty and it’s also just not very accessible. Many of these systems that are even that you might see in other ecosystems where you have a company that says, hey, I’m going to donate some of my own machines time. Some of those systems are just completely air gapped from the Internet, so they’re just not accessible. So this has been a common problem that we’ve had for a number of years. And so we’ve had a lot of different discussions with multiple parties, we’ve had members that have really, we’ve had conversations with, we’ve had some stopgap solutions in a few different areas, but never have we really been able to crack the full problem of how can we provide infrastructure that that’s accessible for any open source project that can support the s390x mainframe platform, both on the Z Linux as well as the Z OS side of things. So this is really the culmination of all of that coming together over time. And it was also just kind of a bit of a new thing because there’s never really been a program like this ever before. I’m sure a lot of people have tried to do this and maybe stalled out at different points along the way, but to the point where we’re able to actually make this sort of offering is something that literally has been years and years in the making and it’s a big thanks to Broadcom, Converged Technologies, IBM, a number of our other members who’ve contributed in one way or another to either helping make the hardware available, contributing software to be able to use it, resources, time. This is truly the definition of a community effort.
Swapnil Bhartiya: What does it really mean for of course those who target, who want to build for Mainframe, what challenges that is solved for them, what opportunity it opens for them. And of course we can also talk about how they can access it and who can access it.
John Mertic: Sure. So mainframe is a different architecture, so it’s not like x86, it’s not like arm or defnothers. There’s definitely different underlying things just from an architectural standpoint. And in addition the hardware has a number of different features and functionalities that are very unique to it and very custom tailored processes and things like that and adding another layer on that. While on the Linux side of things, yes, the Linuxes you would get exposed to on a mainframe is very similar if not almost identical to what you’d get exposed to on an X86 or an arm or other box. Z OS is an entirely different animal. Like it’s really very unique in a lot of different ways of how you interact, how you work with it, how you help support it from an operating system point of view. So for developers this has been a huge challenge. It’s like okay, we have demands from mainframe customers of hey, I’d love to run this open source package, I would love to run this open source code. We don’t even have a way to even test. So they just kind of have to do guessing or what we’ve seen in the past is, you know, folks from an IBM or Broadcom or Rocket software, others have just sort of done the porting work and upstream named it back to the community in many cases irregardless, it’s just never been a very efficient method and you’re also sort of creating kind of a little bit of a bottleneck of hey, there’s only a couple people, specialties and companies especially to make this happen and it’s really made it difficult for the larger ecosystem to take off. In addition, our own hosted projects have suffered from that as they’ve came to us and said hey, we’d love to have mainframe infrastructure, hey, we have a mentee that we’d like to get mainframe infrastructure for. We’ve just not had a very good path to help provide that whatsoever. So this here kind of really opens all of that up and really what it is is you know, we have our, you know, interested parties who are open source project maintainers. They can come apply at the Open Mainframe Project Technical Advisory committee. They look at the applications, they sort of make sure that it hits the core requirements which is really this is an open source project, this isn’t commercial work, this is truly designed to be an open source project and it’s designed to really benefit the mainframe community as a whole and it’s really designed to add value for growing the mainframe ecosystem. The groups looks at that and then says hey, this looks good and then refers them on, then passes them on to the backend teams. And we have folks at IBM that are really helping do the work of getting these projects set up with instances and things like that. And the good thing is if you’re an open source project, you’re like, I don’t know what to do, even ask for. Just there’s a standard offering that’s right out of the gate. It’s like, hey, here’s what a standard, you know, Z Linux image. This is what a standard Z OS image. Here you go, ready to go. So you don’t have to know of like I need this many IPLs and this much disk space and this, that and the other thing. It’s a very standard offering that’s really approachable Anybody. We do also have some flexibility that if projects have very specific requirements, we’re also able to accommodate those. So it might be specific underlying hardware, specific underlying systems, other things like that. There is some flexibility in that way and that’s a real benefit that we have through a lot of the partnerships that we’ve built here. So it’s very simple. Apply to the tac. Tac looks at it, says yes, no, and helps move you along to the next step of the journey.
