Author: Alan Shimel

IBM recently GA’ed perhaps its biggest zSystems mainframe release in the last decade or longer. The z14 mainframe from IBM has also been designed from the ground up to be DevOps-empowered. From APIs and other integrations to language support, the z14 is IBM’s most advanced mainframe ever. I had a chance to speak with Hayden Lindsey, VP of DevOps Enterprise Systems for IBM, about the z14.

Hayden gives us the low-down on how the z14 has been designed to work in hybrid environments. It also works well with modern CI/CD tools such as Urban Code and others. Additionally, APIs give it the ability to integrate with many different environments, tools and apps. It is truly a mainframe for today and beyond. Hayden does a great job of describing all of this.

As usual, the streaming audio of our conversation is below, followed by the transcript of our conversation. Enjoy!

Audio

Transcript

[su_pullquote align=”right”] [su_custom_gallery source=”media: 117632″ width=”130″ height=”130″] Hayden Lindsey

Hayden Lindsey, Vice President & Distinguished Engineer – DevOps for Enterprise Systems. Hayden is responsible for the development, product management, sales enablement, and services of this portfolio. He has worked at IBM for 28 years and has held a variety of positions, including software developer, product architect, product manager, co-leader of the Eclipse tools platform, and director of the WebSphere Studio/Rational modeling and construction tools. He has a bachelor’s degree in Mathematical Sciences from the University of North Carolina at Chapel Hill.[/su_pullquote]

Alan Shimel: Hi, everyone, this is Alan Shimel, devopsy.kinsta.cloud, here for another DevOps Chat. Our guest on this episode of the chat is a good friend of ours, Hayden Lindsey, VP, Enterprise DevOps at IBM. Hi, Hayden, welcome to DevOps Chat.

Hayden Lindsey: Hi, Alan. Good to be with you again.

Shimel: Good to have you here. Hayden, it’s a big day—big day for the folks over at the IBM Enterprise team. It’s the general availability of z14 today, isn’t it?

Lindsey: It sure is. Yes, we announced this back in the middle of July and we’ve now been making our products available. A lot of mine were available the latter part of last week, and now the box itself, GA, today. So—a big day.

Shimel: Fantastic. And Hayden, just—you know, our audience being so focused on DevOps, they may not realize the full scale of what the z14 release is about. What it is, it’s a combination of hardware—as you mentioned, a new box. There’s a ton of software, there’s a tone of integration—it’s just chockfull of all kinds of goodies. Can you give us kind of a—at a high level, the scope of z14?

Lindsey: Sure. And I think what you see is the result of pulling software, both middleware and tools, from the various divisions that we used to have like Rational and WebSphere and moving those into a single z or mainframe focused brand.

And so, in contrast to most prior hardware releases where you’d be hearing about the speeds and feeds and, you know, the z13 had the fastest processor of any box on the planet—this time, the focus is more on the entire solution that involves both hardware as well as software and tools. And so I think—we worked on that, we did design thinking, we had sponsor users, we did the whole nine yards form the hardware up through the operating system, middleware, and tools, and I think that made a big difference.

And so this time, for example, we have something called pervasive encryption. We all know how important security is, we all hear about these breaches that are happening—Experion being the most, or Equifax being the latest one. And, of all of the records that have been stolen, and I think it’s something like 9,000,000,000 since 2013—only 4 percent are encrypted. And this new mainframe lets you encrypt 100 percent of the data with no impact to performance. So, that’s just—that’s one example. Clearly, there’s hardware stuff there, but there’s also middleware that had to be involved in making this solution. So, that’s an example—it’s really cool.

Shimel: That’s actually a big example, Hayden, because you know what? Encryption has been around—I mean, I spent 20 years in the security space and encryption has been around for most of that. The problem was the hit you took on performance when you encrypted and then had to decrypt and re-encrypt just the latency and—

Lindsey: And the programmer had to do it. [Laughter] Or the database administrator had to set it up for the database, but then what about when it was being used in the application? And so that was—this is what is being solved, where you don’t have the time inhibitor of people doing it, and then you also don’t have the performance impact where it makes a system unusable.

So, I think this could be really groundbreaking for the companies and the government agencies around the world that deal with sensitive data.

Shimel: Sure—and we all do now. Hayden, you said something else, though, that I wanted to key in on, and that is sort of moving away from the speeds and feeds, if you will, metrics of—you know, to measure the value of anything. And I worked with a woman once, and she used to call that bag diving. Right? You know, when you gotta start getting into the speeds and feeds to show the value of your product, you’re diving into your bag instead of talking the true value.

Lindsey: Right.

Shimel: So, I’m happy to see that that’s kind of the path that IBM is taking, talking about the value, but I also think it’s a result, as you said, of so much being added into here, right, from the design thinking around the entire release, if you will, to the integration with things like cloud, with the DevOps entire tool set and methodologies that IBM and others have developed.

This is so much more than just, “Oh, we put—this is the next iPhone with the faster processor.” This isn’t the next z with the faster processor, it’s much more than that. As I think I—maybe you said it earlier or maybe we talked about it off-mic. This is probably the most substantive release for z in at least a decade.

Lindsey: I think it is, and of course, the market will tell us if our aspirations come to bear or not, but I certainly think the encryption element is just—well, I think it’s phenomenal, and it could be very, very, very significant for virtually any company out there.

And then we’ve got more on machine learning and analytics so that you can do in transaction analytics rather than doing it after the fact and having to try to back out or just write off fraudulent transactions. And, of course, we’ve done a lot with our DevOps portfolio compilers. But I really do think, maybe, since we put Linux on the platform, it’s that significant.

Shimel: Wow. Huge. So, Hayden, we are about DevOps here, so let’s focus in a little bit. I mean, one of the nice things is, I’ve seen some briefings around how you can use the new z14s with the IBM Enterprise DevOps solution and really sort of leverage—I don’t wanna use the term bimodal multi-speed nonsense. Leverage integrating your z system with cloud systems into one coherent business strategy. How’s that sound? [Laughter] So, talk about it.

Lindsey: Yeah, so one of the three themes for the release is open and connected, and that obviously ties into a hybrid cloud strategy. It ties into, you know, a digital economy or digital transformation. It ties into API economy—whichever angle you want to take, you have to be able to make the data that resides on the mainframe, which is estimated that roughly 80 percent of the Enterprise data in the world resides on and is managed by the mainframe, and about 70 percent of the business transactions.

And so, if you want to, though, make that available to your new iPhone X app, as we saw introduced yesterday—you know, or 10 or whatever they’re gonna call it.

Shimel: Mm-hmm. Yeah, they’re calling it 10.

Lindsey: They’re calling it 10 using a Roman numeral.

Shimel: Yeah.

Lindsey: But, you know, if you’re gonna do that, you have to make it easy to access the data and transactions, find those transactions, understand them, because very likely, they were written by somebody who’s not around any more or had moved on to other jobs. And so, you need to have tools that make API enabling and opening up and connecting to the mainframe easy and efficient, and just as easy, just as efficient as you would if you were opening up something on the Linux platform—although, of course, you have Linux on z—or a Windows box or anything else. So, that’s—it’s essential that we allow z to participate in the broader digital economy, API economy, and that’s what my team has been focusing on, along with other teams in IBM.

Shimel: Sure. So, now, Hayden, in that same vein, you know, there’s still that mainframe attitude where, you know, we don’t open our mainframe up to all of these APIs, and maybe we don’t want to hook into our systems of engagement, you know, up in the cloud or what have you.

For clients listening out there—and a lot of our listeners are Enterprise class, they work in Enterprise class organizations—how important is it that they really consider opening up, you know, their mainframes to this API economy, as you called it?

Lindsey: Well, I think increasingly, certainly, the IT and line of business executives, the clients with whom I work, are realizing they have tremendous assets that they could leverage if they were to make them available. Nobody—at least no companies I know of, or government agencies, have the resources to just recreate everything from scratch. And, I think if you look back at what the analysts were saying back in the SOA era, they said, you know, it was five times faster to create a service leveraging something that exists that already does the same thing or something very similar than it is to create it from scratch—not to mention the fact that, if it already works, you’re likely to have higher quality as well. And I think the exact same analysis would prove to be true now that we’re talking more about RESTful APIs.

So, I think more and more clients realize they need to expose that value. The key is, just, it’s sometimes daunting when you’ve got—you know, the estimate is like 250,000,000,000 lines of COBOL, not to mention PL/I and Assembler and 4GLs and so forth. It’s getting your arms around applications that were probably created in a more monolithic, less well-structured fashion. You need to do refactoring. You need to discover what is there, and then put together a plan for doing that API enablement.

But I think people know they need to do it, it’s just, sometimes it’s viewed as a daunting task and they have to figure out how to get started, and we have something called the digital transformation model that we’re rolling out to help our clients put a plan in place.

Shimel: Got it, and I agree. So, you’re just setting me up for my next things here, but—so, that’s the biggest question we hear at devopsy.kinsta.cloud, Hayden, which is—hey, this is all great. How do we get going? How do we roll this? How do I get executive buy-in? How do we—is there a one, two, three? Is there a blueprint I can follow.

So, z14, let me throw it right back to you—where do we start? How do they get moving with this? How do they—and specifically, how do they start incorporating the DevOps and API stuff into this?

Lindsey: Yeah, it’s the question that I hear from virtually every customer I talk to, and I’m getting ready to do a tour in Europe next week, and I’m sure it will come up in every single meeting. And the fact is, every customer is different, so there’s no single answer. And, within any given client, they have applications that span the gamut from nearing end of life to ones that are under active or brand new development. And so, what we’ve—but, of course, there are patterns and similarities across clients and application types, and there are best practices that we have gleaned by working with clients through the years, and that’s where I mentioned this thing called the digital transformation model.

We’ve shifted to this way of working with our clients to focus on applications rather than one answer for the Enterprise—hey, you should start with deployment automation, or you should start with test automation. That answer should not be an Enterprise answer, it should be for this suite of applications because of what their current state is, the as-is state, and what you desire the to-be state to be. If you just need something to run efficiently and you’re gonna replace it in a few years or retire it, all you want it to do is run efficiently.

So, use our latest compilers, but you know, the fact of the matter is, putting in a fully automated pipeline is likely a wasteful effort. Whereas, on the other end of the spectrum, for the things under very active development, that’s where you want to do DevOps in its fullest, and really, it’s just a matter of what steps do you take to get there.

And so, this model that we’ve come up with has four levels—the first one I described about, just do your compiling, is run—just run it. Just run it efficiently. The next is maintain, the next is expose, the next is evolve. So—I mean, and I don’t know if we’ve got the best names in the world, but the fact is, you really need to think about, for the business, what degree of DevOps adoption do you need for a given application or application suite. So, that’s the more, I think, realistic answer that we’ve been using with clients this year and a tool that we’re using that’s different from what we’ve done in prior years.

Shimel: Excellent. So, Hayden, I mean, just to illustrate this, I remember back at Interconnect last year, you know, we had the whole little DevOps TV booth set up, and I was interviewing some IBM clients. And one of them, it was a bank, and I forget if they were from Poland or the Netherlands now, or maybe even Australia. They were IBM’s z System customer and they had just updated, I wanna say, from COBOL 4 to 5—maybe it was 5 to 6, I don’t remember now.

But they had just updated their version of COBOL, and it made a huge, huge difference for them in terms of performance and—

Lindsey: Yeah. Yeah, exactly. You know, we’ve—one of the things that our compiler team does every single release, and this starts from the very earliest days of, we were just imagining what the new mainframe or new power platform is going to be. We were working hand in hand with the hardware folks.

And, you know, with COBOL specifically, where something like 70 percent of the world’s business transactions are running COBOL, obviously, performance does matter, and so this is where you can get into a little bit of feeds and speeds. Mainly because, not because—some of the numbers aren’t eye popping, but it’s because you can be more responsive for your clients. But with some compute intensive apps, they run 16 times—not percent, times—faster on the latest compiler, which is 6.2, versus 4.2, which may be where that client was coming from. So, that’s skipping the version 5 things, but we still have the majority of our clients on the 4.2 version.

So, it’s tremendously faster, but this is by doing this joint development. You know, if you put new instructions or new vector support in the hardware and the compiler doesn’t generate these new instructions, well, then, obviously, no client is gonna be able to take advantage of it unless they want to write their own assembler. So, you know, our compilers are delivering tremendous value, and I gave you a COBOL example, but you get similar results with PL/I and with our automatic binary optimizer that basically just does the, tries to do the work that the compilers do, but just working on binaries rather than having you go back to the source.

So, it’s a big deal. You know, and the less—it’s maybe a bit more mundane. Yeah, the stuff runs faster and, as you said, diving into the bag. [Laughter] But, you know, when you’ve got numbers like 16X or 5X or what have you, that’s huge.

Shimel: It sure is, it sure is. So, Hayden, you know, I mentioned Interconnect, too, and as we’re running low here on time, I just wanted to quickly prompt you on that and let our listeners now.

So, this year, Interconnect is part of an even bigger kind of unified conference that IBM has put together called Think—Think 2018, I believe it’s called, and if I’m not mistaken, it’s March.

Lindsey: Yes.

Shimel: Go ahead.

Lindsey: Well, you might see a trend here. [Laughter] Because, you know, Interconnect was the combination of Impact and Innovate. Innovate—I think that’s the name it was. I’ve kinda forgotten, but anyway, we had a Rational conference, we had a WebSphere conference, we had IoT stuff and all that. And now we combine those into Interconnect—at least the WebSphere and Rational stuff. And it got bigger, and now we’re doing something that is more or less IBM-wide and it’s called Think. It’s gonna be the week of March the 19th at the Mandalay Bay in Vegas, so still in Vegas like most of our other conferences have been of late.

And so, I think there’s pros and cons to having a more broad, where we can tell the full IBM story, and that’s the focus, and you’ve got SMEs from all the different areas of the country, be it cloud, cognitive, Zpower, what have you. But I think that’s great, but because you do have those experts, and we will have the breakout sessions or keynote sessions on specific areas, be it DevOps or z or compilers or deployment automation like UrbanCode.

So, there will be still those specialty sessions, and yet you get the big picture. And the one thing that I would like to ask your listeners is, the call for papers is open, and you know, we are shooting for having 100 percent of the talks in the various tracks either be a customer or partner only, or being customer and/or partner jointly with an IBM-er, but we’re just really trying to get away from IBM-er talking head telling a story without a partner or a customer there saying what they’ve done in real life with a given solution in a given area.

So, we’d really love to have as many partner and client nominations as possible.

Shimel: I love that idea. You know, I shared a panel the last two years, Hayden, on Dev—I think they were on, well, it was on DevOps and then SecOps. But, you know, I tried really hard to balance it with—yeah, we had one person from IBM, but we also had an analyst and then we’d have a partner and then a practitioner. And I think that sort of balance gives it, it makes it—

Lindsey: It adds some legitimacy to the whole thing and different perspectives.

Shimel: Yep. Just real quick, for anyone listening, if you go to IBM.com/events/Think, you can get all the scoop there and follow the links to submit. And as Hayden said, if you’ve got something to say and share with people, it—having been to three or four of these, or more, it’s a great platform. So, I’d encourage everyone who would like to participate to do so. And even if you don’t want to speak, but if you want to go really understand what’s happening, whether it be with Watson or z Systems or hybrid cloud or Bluemix or any of those things. You know, if any of those are important to you, I think it’s a great show to attend.

Hayden, we’re about out of time—we’re actually over time, but that’s okay. Thanks so much for taking the time on what is a busy day with the general GA of z14. Best of luck with it. Let’s check in, I don’t know, maybe before the end of the year, and see how—you know, what kind of market reception a lot of these features and capabilities are receiving and we’ll see where the rubber is meeting the road.

Lindsey: Sounds good, Alan, and thanks, again, for your time.

Shimel: Thank you for your time. Hayden Lindsey, VP, Enterprise DevOps with IBM on GA Day for z Systems z14. Thanks, Hayden, have a great.

Lindsey: Okay. Bye for now.

Shimel: This is Alan—thank you. This is Alan Shimel for devopsy.kinsta.cloud. We’ll see you soon on another DevOps Chat.

— Alan Shimel