Picture in a mainframe
Jonni Bidwell wants to know if he could install Linux on the mainframe at Future Towers. The Open Mainframe Project’s John Mertic has all the answers.
Jonni Bidwell wants to know if he could install Linux on the mainframe at Linux
Format Towers. The Open Mainframe Project’s John Mertic has all the answers.
on mainframe users “You have a community that’s always been very much open source and collaborative-minded.”
John Mertic is Director of Program Management for the Linux Foundation’s Open Mainframe Project
(www.openmainframeproject.org), an effort to bring open source knowledge, and of course Linux, to mainframe computing. You may be forgiven for thinking mainframe machines were all clunky relics running obscure COBOL code, but the modern mainframe is a thing of technical beauty. While the cloud offers elasticity, portability and spares you the woes of managing your own infrastructure, mainframe(s) offer resilience, security and performance.
The project uses Zowe, the first open source software to run on IBM’S Z/OS – and yes, you can now run Ubuntu on your mainframe. Besides being a mainframe man, John is also Program Manager for a number of other Linux Foundation projects, including: Open Data Platform
(www.odpi.org), a non-profit dedicated to standardising the big data ecosystem; the R Consortium; and the Academy Software Foundation (www.aswf.io), whose goal is to further the role of open source software in creative industries. He was good enough to spend some time chatting to us at the Linux Foundation’s Open Source Summit back in October 2018.
Linux format: Mainframes always seem inaccessible to the lay person. They’re bulky and expensive and it seems like they can only be operated by people with crazy hair (shouldn’t throw stones?–ed) in white coats. Is this changing?
John mertic: A couple of years ago there was a student who bought a mainframe off ebay. He had it in his parents’ basement1 and he figured out how to get all the software and everything running on it. The mainframe community was blown away – he got to bring it to conferences and everything. I think IBM helped give him a bunch of missing pieces that he needed; I think he works for IBM now. I guess if we classically look at where you’d see mainframes in a historical context, my kids would tell me it’s the spaceship earth ride at Epcot [at Walt Disney World in Florida]. You’re riding by and you see the mainframe setup and the dudes with big hair. It’s kind of a shame
– if you search Google for ‘mainframe’ you see so much of that being the context. I think the mainframe community’s quite sensitive over that, but I’m a big fan of how we look at the legacy of these things.
When the current generation of mainframes rolled off the line in 1964 they were all designed with security, performance, scalability and availability as the core tenets. And every design decision and every hardware addition builds on those. That’s what’s kept that ecosystem rolling, even up to the boxes today. The latest one actually fits into a standard rackmount unit; before, they used to be a lot bigger. It’s interesting technology, and if you’re running SUSE or Ubuntu on it, you really can’t tell the difference between it and a regular server – it all just works the same.
LXF: There’s a lot of talk at this conference, most of which is very confusing to me, about new technologies like IOT, microservices, fog computing, edge computing. But I think it’s easy to forget we owe a lot to mainframes, and they’re still a big part of computing today. I was talking to the IBM folks last year and they told me that every card transaction and every flight booking goes through a mainframe computer. So even if we don’t see them they’re still very much there, and very much aren’t going anywhere soon, despite often being seen as a relic. What is the Open Mainframe Project all about, and how will it help with this retrospective?
Jm: The Open Mainframe Project is a foundation launched to support the growth in open source projects on mainframe. Very astute, by the way, your knowledge of the mainframe ecosystem, and where we’ve come from. It’s really interesting in an open source context. If you want to trace back the roots of open source back to predecessor organisations, you’ll find it all traces back to 1955 and SHARE. SHARE is a mainframe users’ group; it still runs today, but it was the concept that all these mainframe users, programmers, administrators, systems people would all get together and share source code. Back then they were using things like microfiche and tape and things like that, but it was that same idea of sharing (that we have today).
So you have a community that’s always been very much open source and collaborative-minded, but never used that terminology. That’s always been a root part of this ecosystem. The open source movement came alongside of this. So this project is about how can we pull all of those different collaboration efforts together, so there’s a natural home where you would go to get support. Whether you’re an open source project and wanting to support the mainframe, or if you’re already on the mainframe and you want to support open source, that’s what we’re here to do.
LXF: Which industry partners are you involved with?
Jm: Lots of organisations – obviously IBM, they’re the primary mainframe vendor, I think they have something like 90 per cent of the market, so that’s the hardware we focus on predominantly. But we also
have Linux vendors such as Canonical and SUSE involved. Then there’s application vendors like CA, Rocket Software and Docker. We also have universities and IBM systems partners involved.
There are also just general end-users of mainframe in the community. I don’t know if ADP is well-known in Europe, but it’s the largest payroll processor in the US. Most Americans and Canadians get a pay cheque or a W2 from ADP, and it’s all done from Linux on a mainframe. It’s a pretty wide spectrum, it’s also a wide spectrum of generations. So you have 70-year-old systems programmers, you have middle career people, you have new developers. You have all these intermixed dimensions that are coming together.
LXF: Banks in the UK are rubbish. I mean, they’re probably rubbish everywhere but in the past few years we’ve had some pretty major outages: people’s pay not being credited, people not being able to access funds, mortgage payments bouncing. And a lot of the time this is blamed on failed mainframe updates. The feeling I get is that we’re losing expertise here, as people that understand these things are retiring and, well, expiring. I mean, who learns COBOL nowadays? Can the Open Mainframe Project ensure this ancient knowledge is taught to a new generation, so that our funds are not at the mercy of fatfingered updates?
Jm: There is a generational gap, and I think that’s one thing the mainframe industry is forward-thinking about. It’s a 54-plusyear-old industry, how’s it going to get through the next 54 years? It’s going to be through getting that next generation involved. There’s a number of different efforts that are underway. I think one piece is that the drive of Linux and open source and mainframes already having that commonality between it.
Frankly speaking, mainframe can’t be replaced and not just because of the applications that are already locked into it, more because of what it can provide. If you’re an organisation and you need an application that needs the dials of security, performance and reliability all turned up to 11, then no cloud system can do that for you, no distributed system can do that for you, only the mainframe can do it. That’s why these organisations are centred on the fastest commercial processors on the planet running inside mainframes.
Being able to have encryption as your data is read into memory and moves back onto the disk – hardware-based encryption – that’s only on mainframe. All of these innovations happen here, resource-intensive, high transaction computing happens here. It’s not going away and it’s not going away because there’s nothing else that can replace it. So the big question is how do we get that next generation engaged, but then also how can all of these applications that are built on mainframe, that are perfect for it, that are completely designed for it, how can the data generated from them be taken advantage of in the rest of the organisation? Because, and I’m sure the UK probably sees the same thing, the mainframe group is always silo-ed off away from the rest of IT, and that becomes a huge problem.
So on one hand I think the emergence of Linux on the platform, the growth of Linux on the platform is a huge piece there. We’re already seeing projects, we’ve worked with Kubernetes, Cloud Foundry, Openstack, they’re all starting to pull together. Organisations can, from a devops perspective, manage all of that infrastructure in unison, versus having specialised management. The other project that we’ve been working on, that we launched in summer 2018, is a project called Zowe (www. openmainframeproject.org/projects/ zowe) and what that’s focused on is the z/ OS side of mainframes. That’s sort of the legacy operating system there.
Before Zowe, the only way you’d interact with it was over what they called 3270 terminals, which in layman’s terms are green screens. What that does is it puts a set of REST APIS on top of that, so that any modern application can easily tie
in to all the mainframe data. Additionally it adds a command line interface, just like what you have if you SSH-ED into a regular Linux box. There’s a web UI portion there too, so you can basically build desktop apps in a browser that ties right into the mainframe. But it’s all Electron/ Node.js built, so you can easily integrate different sources as well. That project has the opportunity to revolutionise, so that (for) all these legacy Z/OS apps that are out there, and are going to be there forever, at least there’s a way that data can be migrated back into the rest of the infrastructure. So now that’s helping to make more commonality amongst the toolsets that people are using.
So we have those two focuses, and I think the third focus that we’re working on is our internship program. They actually presented here at the summit today – we brought the gang along. As what we’re really trying to say is “Hey students, you’re into open source, there’s a lot of open source here on mainframe, there’s opportunities. Come spend the summer with us and work on it”. And they make great contributions – we’ve had students that have made contributions to Hyperledger, Kubernetes, blockchain. We had one student that ported Alpine Linux over to the platform.
Then all of those students then come out of this program, and they’re already connected to mainframes, they’re already familiar with the technology, they see the great opportunity for, frankly, very highpaying jobs. They’re some of the best-paid new graduate Computer Science jobs out there. And (through the program) they already have a foot in the door and a career path. Then that’s helping all those companies – going back to your point before – that are trying to backfill all those positions resulting from the retirees.
LXF: Yep, occasionally you see job adverts for VAX experts. Does anyone understand that any more? I mean, I once saw a VAX machine – it was being used as a coffee table. Very sturdy. Not sure if it still booted.
Jm: Well, VAX is probably a whole different ball of wax there. But now there’s a lot of these technologies out there that are going to be around forever and I think that’s an interesting challenge to the open source community. What can open source do to unify all these? Maybe we’re taking the lead here with our Project.
LXF: With Openstack and Cloud Foundry and Kubernetes and such, you can run all these at home on a bunch of virtual machines. It might not run well, but for illustrative purposes and to get a handle on them, you can do that. You can’t very easily get an old mainframe and plug it into a normal power supply.
Jm: The interesting thing is, by having that ability where the software layer is common across all those architectures, now you can use the architecture that best suits what you’re trying to accomplish. I mean, you and I are lucky to live in an age of technology. That wouldn’t be the case 20, 30 years ago – you’d get a handful of vendors, and once you purchase your hardware from them you’re stuck in their ecosystem, you’re not getting out. Now you have Linux as this commonality across all of these architectures, and then you have the value-adds that are popping on there, Kubernetes and all these different technologies that are all ubiquitous across all of this.
Now you can say “You know what, if I have an application here that I just need ridiculous performance on, I have high security needs, then I can flip it over and I can run it on the mainframe. What if I need to run at the edge? OK, fine, I’ll push it onto an Arm box. And if I need some elasticity on the platform, then it can go in the cloud”. You have all those choices. We live in an age where we have a gigantic menu of computing choices, which is fantastic. It’s a great thing for organisations to take advantage of, because they’re becoming more and more heterogeneous in their environments.
LXF: Apart from my two examples earlier, which I practised in the bathroom mirror this morning…
Jm: Ya nailed it, buddy.
LXF: Thanks. So apart from those, I also learned a little bit about pervasive encryption. That’s something that’s peculiar to mainframes, it would be frightfully complicated to cloud-ify it. Can you tell our readers something about it?
Jm: Pervasive encryption is an IBM technology on the latest z14 boxes. The nutshell is, as data is hitting memory, it is automatically hardware-level encrypted, not at software level. It’s encrypted before it’s even hitting the disk, so that’s an additional layer of security to that data there. Mainframes are quite unique because mainframes have never been hacked – they’re the most secure things on the planet for a reason. And pervasive encryption just continues to add to that,
the need for training “There is a generational gap, and I think that’s one thing the industry is forward-thinking about.”
so if you have those sorts of requirements then that’s what you need. We’re seeing other technologies take advantage of this too; IBM is working on a secure containers offering that’s using KATA containers. We had a student this summer leveraging that same technology, again to make this platform as secure as it needs to be.
LXF: I always like to pore over specifications and compare them to this pitiful machine from 2011 that I do all my work on. So how powerful are these machines, really?
Jm: I think the z14s have 5.2GHZ processors (they do indeed – Ed), and they have about a hundred of them. They don’t have what we would call RAM, they call it RAIM (Random Arrays of Independent Memory) – it’s fault-tolerant and they can have up to 32TB. You can hot-swap any single piece of a mainframe out while it’s running and it just keeps on ticking. There are pictures, I think from the tsunami in Japan in 2011, where a building collapsed and you saw a mainframe toppled over in the rubble, but it was still running.
LXF: What about other applications of mainframes?
Jm: One of the more interesting workloads that it’s really well suited for is blockchain
(see box on page 40). Blockchain is all about ledger transactions. Plasticbank (www.plasticbank.com) has really taken the lead here. They’re an organisation that’s focused on reducing ocean waste. They’re out there using Linux on a mainframe for doing all the co-ordination between collecting plastics from third world countries. That’s the only way they could do it at scale.
A lot of the traditional applications are still on the rise. Even more banks and financial companies are setting up, especially in South East Asia, China has a big push too. The saying my mainframe colleagues like to quote is that the cloud could go down, the cloud could disappear. (If that happened) it would be tough for us, but we would get by. There would be a lot of inconveniences in life, but we could get by. If the mainframe went away, if all the mainframes in the world shut off, society would plunge into chaos. You wouldn’t be able to access money, and all the derivative chaos that comes from that.
LXF: What developments are we going to see coming soon from the Open Mainframe Project?
Jm: I think we’re working on a couple of different fronts. One is continuing to foster out collaboration efforts within the mainframe ecosystem. Zowe came from a collaboration between IBM, Rocket Software and CA Technologies. They all contributed pieces and said, let’s bring this out as open source. There are some other projects we’re working on in the wings. ZVM is the hypervisor for mainframes, analogous to KVM. One project currently in incubation is to make it trivial to connect Openstack or Openshift, or something like that, to leverage ZVM as a hypervisor.
Another thing we’re just getting more formalised is supporting projects that might want to support a mainframe. The biggest challenge for a lot of those projects is getting hold of a mainframe to test it on, so we’re providing mainframe infrastructure for building, CI and testing build environments they can use. We’re also helping connect them with expertise if they run into problems and showcasing their stories. It’s our goal that, over time, mainframes should be so ubiquitous within computing that it almost doesn’t need its own group to promote it. It’s not like there’s an x86 Foundation, right?