IBM z17: Doing More at the Core - Six Five On The Road

Is the mainframe still the undisputed king of reliability and security? 🤔

‍

While the cloud and AI may dominate the headlines, the mainframe remains the backbone of the global economy, processing trillions of transactions daily. At the z17 launch, Patrick Moorhead and Daniel Newman spoke with IBM Z’s Tina Tarquinio, VP of Product IBM Z & LinuxONE and Chris Berry, Distinguished Engineer, about how this new generation is pushing the boundaries of reliability, security, and AI integration!

‍

Key takeaways include:

‍

🔹Built for the Unbreakable: z17 is engineered to handle the most demanding workloads with performance, scale, and security. We're talking about systems that underpin global finance, healthcare, and critical infrastructure.

‍

🔹AI Deeply Integrated: IBM is embedding AI at the core of z17, enabling clients to leverage powerful new capabilities for fraud detection, risk assessment, and real-time insights, taking advantage of the platform's inherent low latency.

‍

🔹Next-Level Resilience: z17 enhances the mainframe's legendary reliability with advancements in core recovery and other hardware-level innovations, ensuring near-continuous availability for mission-critical applications.

‍

🔹Modernizing for the Future: IBM is committed to making the mainframe more accessible and user-friendly for a new generation of developers, while also leveraging open technologies to integrate it seamlessly into modern IT environments.

‍

Learn more at IBM.

‍

Watch the full video at Six Five Media, and be sure to subscribe to our YouTube channel, so you never miss an episode.

Or listen to the audio:

‍

Disclaimer: Six Five On The Road is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.

Patrick Moorhead: The Six Five is On The Road here on the IBM z17 launch day. Congratulations everybody at IBM and your clients. We are here at the One Madison, IBM's brand new building here in New York City. It is so new. I'm pretty sure that I smell paint. I don't know. New carpet smell. Dan.

‍

Daniel Newman: Yeah, you gotta love the energy walking into this place. Been here on investor day, which we'll talk a little bit about now back here for the Z launch. And Pat, it's just great to see the kind of transformation that IBM is going through. This launch is a big moment for the company, but it's really more about the whole company being on this theme, this hybrid cloud and AI theme and just seeing it continue. And that's probably a lot of the reason, the success. You've seen it in the share price, you've seen it in the business, you've seen it in the growth company's really been able to continue its path forward under Arvin's leadership.

‍

Patrick Moorhead: Yeah, it was a great conversation with Z General Manager Ross Mauri.

‍

Daniel Newman: Check it out, link, check that out.

‍

Patrick Moorhead: Yeah, absolutely. But we're here for, I would call it the double click on z17. Right. With Chris and Tina. Welcome to the show.

‍

Tina Tarquinio: Thanks for having us. Longtime listener, first time joiner.

‍

Patrick Moorhead: This Is wonderful. I love that you're hired. No, this is great. We've done a lot of conference zoom calls, not zoom calls, I don't know, video calls together. Going through this is the first time kind of publicly going through this. And I really appreciate all of your time here. So thank you.

‍

Tina Tarquinio: Yeah, we're really thrilled to be here and thank you for your time. I know you got us early and often on z17 on some feedback and we really appreciated your time and industry expertise there.

‍

Patrick Moorhead: I appreciate you letting me into the inner sanctum when you did. So thank you.

‍

Daniel Newman: Wait, did you know something?

‍

Patrick Moorhead: Possibly. 

‍

Daniel Newman: So it was great spending some time. Tina, actually with you, I was here on investor day. One Mad. Great event. It was a great day for the company. I think the market really got a flavor and Z got some real stage time, which I know sometimes it does, sometimes it doesn't, depending on the cycle. When we did talk to Ross, Tina, we talked about how z16 sort of shifted from everything always being a bit of a cycle, quick spike fall to something that was much more sustained throughout z16. I think 17 looks like it may be setting up the same way. But at the event Rick Lewis got on stage, he really shed some light, bringing this to the forefront to talk us through the kind of IBM mainframe program. And why has this been so successful, so important to the company for so long?

‍

Tina Tarquinio: Well, I think when you look back at the enterprise applications that we've really supported, Right. I mean we made a commitment to our clients that they would never have to change their applications to keep running on the mainframe. And we've kept that promise for almost 61 years, which is pretty incredible. And I think the clients are 

‍

Daniel Newman: Since you were 12. 

‍

Tina Tarquinio: I think clients are at the heart of everything we do. So we really employ something called IBM design thinking and it's, I call it, obsessive behavior around getting our client input. So on z17 we have over 2000 hours of direct user research on the program and that's really how we decide what we're going to do, how we're going to do it, how we're going to deploy it. And I think when you're really that obsessive around your client outcomes, that's what happens. And we're a full stack program, right. So chip to ship, as I like to say. So Chris will tell you a little bit about the chip, but we own the full stack and there's really a lot of magic that can happen when you optimize that way.

‍

Patrick Moorhead: Yeah. Isn't it ironic that that's kind of the new thing. Right? That started this new thing that started seven or eight years ago. On an industry basis you've been doing it forever.

‍

Tina Tarquinio: Yeah. I love when they talk about a platform and it's kind of been our approach for a long time now. Yeah.

‍

Patrick Moorhead: In the industry it's kind of. I call it the according effect, where it's like aggregation. There's disaggregation, there's aggregation, but we are definitely in a point of aggregation at this point.

‍

Daniel Newman: So there's like a mainframe version of AI Factory that's.

‍

Patrick Moorhead: There we go, they're doing it. They don't use that term, but it's.

‍

Daniel Newman: Pretty much you jumping on the bandwagon of what gets all the attention. But I think it's a nice call out that they were doing this before it was cool.

‍

Patrick Moorhead: No, it is. So let's do the double click on z17 here. Maybe you can talk about what's new, what's been modernized and bonus points for talking about benefits to clients.

‍

Tina Tarquinio: Sure. So I hope everybody will see, but one of our slogans is you can do more at the core. And so when you do more at the core, you'll have more mips, more capacity, more memory, all that. More at the core.

‍

Patrick Moorhead: Can you explain the core concept to everybody who might not be familiar where that comes from?

‍

Tina Tarquinio: Yeah. So cores sort of have a double meaning. Right. So there's cores on the processor and so you'll be able to do more at that core. More AI acceleration. Like I said, more capacity, more memory, more speed, more IO, all that. But at the core of your business, you can really do more. Take advantage of the investment in the mainframe and the technology. To me, when we first started to talk about the messaging more at the core, I was like, yes, that's absolutely. We want our clients to be able to do more at the core of what matters to them. So we're going to bring them more AI infrastructure, which is really timely. And it's pretty incredible that our teams were thinking about this four or five years ago because that's how long these take to make. We'll introduce a new PCI attached AI accelerator. We'll have new capabilities of the on chip acceleration. We're going to have IO acceleration on the processor and then full stack exploitation around security and generative AI capabilities with our Watsonx platform.

‍

Patrick Moorhead: Yeah, more at the core. I love that.

‍

Tina Tarquinio: Coming to a T shirt near you maybe one day.

‍

Patrick Moorhead: No, it's good. I mean, you even hear when CIOs talk about their applications, our core applications. Right, right, those strategic applications. So listen, I'm not a marketing person, but kudos to the marketing people.

‍

Daniel Newman: What?

‍

Patrick Moorhead: Oh, I was mostly product marketing.

‍

Daniel Newman: Oh, come on, come on. You're a good marketer. You get it. We build products together. Six Five.

‍

Patrick Moorhead: I know.

‍

Daniel Newman: Anyways, but let's talk a little bit more about your double triple entendre. I'm not sure, but there's a lot of things we can do with core, let's just put it that way in this industry. So talk a little bit about optimization of performance and scale. I mean, when we talk to Rossm Chris, he talks about like, you know, 45 billion, I think was the number of transactions or four and a half. Four and a half or four?

‍

Chris Berry: 450 billion.

‍

Daniel Newman: 450, yeah. Transact billion. Yeah.

‍

Patrick Moorhead: I see a lot of billion.

‍

Daniel Newman: Lots and lots of transactions a day. And the thing is we talked a lot about how it's critical. These are not like ihey, I need to get gas in my car, I need to pay the bill or I need to get on the, you know, and book an air flight. All these things are happening. So talk about how you sort of think about that from building these cores for technological advancement.

‍

Chris Berry: So, right. What you have to think about, it's, as Tina says, full stack integration, but it's also the infrastructure baked into the hardware. Right. We have to start with these massive caches. Right. One of the big updates we made on Telum II, 36 megabyte L2 is. Right. Biggest in the industry, 360 megabyte virtual L3. And that is to enable sort of heterogeneous workloads, large databases, lots of transactions to get to those sorts of 450 billion transactions numbers. Right. You can't do that with standard off the shelf products. Right. You have to have something that's custom built, custom design for that whole stack integration for those solutions for the clients.

‍

Patrick Moorhead: I want to hear about Telum II. And I want to hear about Spyre.

‍

Chris Berry: Yes. Okay, we'll start with the core. So more from the core is we really focused on sustainability, performance per watt. We went after a much smaller core, more power efficient. Right. So you increase the number of mips, increase the number of transactions you can handle in a better form factor. Right. It's smaller, it's more performant, it's more effective. And I talked about the caches.

‍

Daniel Newman: We don't need to put the racks in the bottom of the ocean to keep them cool.

‍

Chris Berry: No, correct, exactly.

‍

Patrick Moorhead: Maybe the Antarctic. Not even the Antarctic. So.

‍

Chris Berry: Make it a little cool. A little cool. So we really are improving, focusing on, across all the dimensions. We're focusing on execution of transactions, we're focusing on scale. So that's where the caches come in. Ability to handle a lot all at the same time. We did a big redesign in our I/O subsystem. We onboarded a data processing unit for I/O acceleration. Transformative from a system integration. What that means in terms of what our cards look like and how, how we integrate all that. And then as Tina said, I/O performance, total bandwidth, total capacity, resilience improves when you talk about those sorts of things. And then, yes, now Spyre cards, right? You got room in the I/O drawers for all the Spyre cards for AI compute and AI processing. And you're talking some of the order of two orders of magnitude more AI compute in the system thanks to the combination of the on chip AI accelerator improvements and the Spyre card.

‍

Patrick Moorhead: Different types of workloads. Right? You might have, might be doing the AI on chips for a certain set of workloads and then maybe some other on Spyre.

‍

Chris Berry: Well, it's really dialing it in for what that client workload needs. Right? For a fraud detection banking transaction, you want that on die accelerator, low latency. You don't want to be standing there waiting for your credit card to approve or disprove or reject whatever the word is, your transaction. You want that to be quick, you want it to be within a millisecond. I think our agreements with our clients. But if you want more accuracy, you want to have that extra compute, if you want to get into Gen AI, you want to have those Spyre cards sitting out in that subsystem. So you have that raw AI compute capacity for whatever you want to do with it. And right now we're trying to open up the aperture for what our clients can do. And that's what z17 will enable. They'll be able to go and explore what they can do from an AI perspective in ways that they couldn't before.

‍

Patrick Moorhead: Now that's great. So I'm just asking all the questions, Dan, just to leave no time for you. No, I'm just kidding. I wasn't there. I was at Mobile World Congress, had a great conversation with Vodafone and they talked about quantum safe computing and it's kind of a conundrum, which is we can debate how many years we are out, but the installed base and what you have is going to live for a long time. So regardless of what year and plus question mark, it has to be incorporated in there. Can you talk about how you're incorporating z17? What the latest quantum safe technology, maybe security as a whole. And Dan made a great point. We were talking with Ross just sometimes how we don't fully factor in how important the security element of this is. And it seems to me that you add an AI capability to that, it just needs to be more secure based on what it can do.

‍

Chris Berry: Yes. So I'll talk about it from a hardware perspective, but I also want to take a step back from a historical perspective. You have to remember, I mean IBM Z is the securest platform, right? That is what it's designed for, that's what it's intended for. And we keep moving the ball forward. Right. When you talk about sort of NIST standards for encryption and technology, right. Those are IBM research algorithms that go into deciding what those standards look like. Those are already baked into things that we currently do and we just continue to improve and upgrade that. So. Right. Specifically. So you start with that base and then you go, how do you improve generationally? And we do things like the SMP cabling that we have going from drawer to drawer is now encrypted and quantum safe. And you know, what we store in memory is now encrypted or has been upgraded and is quantum safe. So you continuously improve across every dimension that you can identify where there might be something that yes, to your point, in plus or however many years out into the future it's going to be, you've got to protect against that because you want to maintain the security and the confidence of your clients. And then yes, from an end user perspective, I want to know that my banking transactions are safe. I want to know that my credit card information, my medical information, all of that is secure and protected. And that's all sitting on an IBM Z somewhere.

‍

Tina Tarquinio: Just to add to what Chris said. So not only did we create this technology and use it to protect the infrastructure and then create the algorithms for our clients to deploy. It's a heavy lift. So we created tools that help them figure out where their calls are to create the inventory and to get started, it's important that they get started. So really looking at it holistically like that I think is really what is leading to some of our success to your question earlier.

‍

Daniel Newman: And it's been good from a standpoint. I know this isn't a quantum announcement, but IBM is doing a lot of things to advance the quantum field. Quantum safe, we've heard some of those sort of, oh, here's this new supremacy thing and it's going to break the bitcoin encryption. But what I'm saying is these transaction types of services though, I mean this is going to be one of the attack vectors, is going to be where money.

‍

Patrick Moorhead: Yeah, it'll be the primary attack vector.

‍

Daniel Newman: And so you having, like IBM having the research business and the quantum business and the Z, like kind of, you really do become a bit of a panacea for all the things needed to support.

‍

Tina Tarquinio: We're working super closely with our research organization. I mean they're part of, I think of some of the success of IBM as well. I mean partnering with them on these very far leading technologies. So they get them out front and then we bring them to the product and that's a really great partnership we have. Quantum safe is a perfect example of that.

‍

Daniel Newman: We can keep Z cool by putting it in the front fridge with the quantum.

‍

Patrick Moorhead: Nice try there buddy.

‍

Daniel Newman: I'm building, I'm drawing a picture.

‍

Patrick Moorhead: No, I know, it's cool. Well, it's kind of a refrigerator rack. I got you. You know one thing that I have been observing that I think should make IBM clients sleep well at night is not just that IBM takes a standard off the shelf. You're helping to create the standard, you're researching what goes into that. You're a leader in that community. And I definitely picked that up at MWC talking with, I know it's MWC but you're part of a lot of different standards organizations that are doing the base research.

‍

Tina Tarquinio: Yeah, across the board. I mean in our IO technology we have distinguished engineers on the FICON spec board. Anne Dames, who's our distinguished engineer for this crypto technology was just asked to join the NIST board. So we are really leading the way and all driven by what we want our clients to be able to achieve. So that's what makes it exciting to me.

‍

Daniel Newman: Well one of the things that you want your clients to achieve is having basically perfect reliability. I know there's no such thing as 100, right. It's always some variant of nines.

‍

Chris Berry: However many nine frames you can get..

‍

Daniel Newman: In the mainframe, you want it to be many, many, many nines. So as we sort of wrap up here, Chris, just talk a little bit about how the z17 is enhancing what's already been sort of the industry standard for reliability.

‍

Chris Berry: Yeah. And so, again, to set the stage for everybody, Right. You're talking eight nines of availability. Right. That translates to for over. This machine's running day after day, year after year. Right. Eight nines of availability turns into an hour of downtime out of every 11,000 years. So since the beginning of civilization, one hour of downtime. Think about how often your laptop crashes or you have to reboot your phone. Right. And then put that in. Exactly. So that's the kind of scale we're talking about here, from a z17. So to get that, you have layers and layers from the software, from application layer, operating system, firmware, and yes, down into the hardware. So we're focusing on, at least from my perspective, from hardware, we're working with our technology team trying to identify any potential vectors that could create reliability problems. We're improving what we have from a core recovery from a reliability perspective down on the hardware, to the point where the software doesn't even know that certain things are happening. We just report it back and say, hey, this thing happened. Just so you're aware, we took a core offline, put it back online. Nobody even knows those are the kinds of things. And you layer that across the stack and that's the only real way you get to those eight nines of availability.

‍

Daniel Newman: Chris is also driving a little observability there. You didn't catch it.

‍

Patrick Moorhead: No, I did.

‍

Daniel Newman: Well, good fun, Tina. Chris, we want to thank you so much. Congratulations on the z17 launch. We look forward to sort of following the journey, continuing to have these conversations over the next few years. And of course Pat will start to give you great  advice on z18.

‍

Tina Tarquinio: Yes.

‍

Chris Berry: We look forward to it.

‍

Patrick Moorhead: Oh gosh. Do we talk about a z18? I guess we are allowed to do that now aren’t we?

‍

Tina Tarquinio: Well, we’ll start that next..

‍

Daniel Newman: We’re not really supposed to but we’ll use our 7G phones..

‍

Patrick Moorhead: I did see the slide at Z so.

‍

Tina Tarquinio: Well we’re already working on it, so. 

‍

Patrick Moorhead: Absolutely.

Daniel Newman: On our personal quantum computers that we put in our bags. We’re getting there, we're moving towards the future but for now. Future. z17. Come back and join us again soon. Thank you so much for being a part of The Six Five. We are on the road here at One Madison IBM’s flagship location in New York City for the z17 launch day. We appreciate you being a part of our community, subscribe and watch all our coverage here from the event. And of course all of our Six Five content. We appreciate your being a part of our community, but we gotta say goodbye. See you all later.

‍