Efficiency Unlocked: Powering the Future with IBM LinuxONE - Six Five Virtual Webcast

How are enterprises optimizing performance and sustainability while facing ongoing operational and environmental pressures?

‍

Host Patrick Moorhead, Founder, CEO, and Chief Analyst at Moor Insights & Strategy, is joined by IBM’s Marcel Mitran, IBM Fellow & CTO, IBM LinuxONE, for a conversation on optimizing performance, reducing costs, and improving sustainability in enterprise IT environments. They highlight key strategies for improving IT efficiency, focusing on technologies and operational models that help businesses meet modern challenges while supporting long-term business and environmental goals.

‍

Key Takeaways Include:

‍

🔹The critical importance of IT efficiency: Why efficient IT infrastructure is essential to enterprise success and business agility today.

‍

🔹Strategies to optimize IT operations: Practical approaches, tools, and technologies for boosting efficiency and reducing total cost of ownership.

‍

🔹Sustainability as a key consideration: How environmental considerations are now integral to designing efficient, future-ready IT systems.

‍

Learn more at IBM.

‍

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

‍

Or listen to the audio here:

‍

Disclaimer: Six Five Virtual Webcast 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: Six Five Podcast is back and we are talking about IBM Linux One. There have been a lot of announcements coming out and more insights and strategy has been tracking it for I think 14 years, which has been, which has been pretty, pretty awesome to do. We track all forms of compute and solutions infrastructure and I just love rolling into New York and discussing this and it's my pleasure to introduce Marcel from IBM. Marcel, how are you doing?

‍

Marcel Mitran: I'm great, thanks Pat. Happy to be here. Excited?

‍

Patrick Moorhead: Yeah, totally. It just seems like the rate of change increases every single year and it's not our imagination. And I'm sure you're seeing this from your clients. There's a lot of things they ask, right? They want to lower costs, they want to drive revenue, they want to accelerate everything that they're doing, even operationally. One of the things I wanted to hone in on in this conversation is is it efficient? So let's talk about in relation to your clients, how are they expressing the need for IT efficiency in this day and age?

‍

Marcel Mitran: I mean, I think it's on multiple levels, right. It's the reality of the bottom line. Right. And just saving cost on power and data center footprint. And it goes beyond that in the extreme. It's now, how do we avoid having to break ground and start, you know, build a new data center? And we're actually seeing that becoming more and more common now as there's no more power, there's no more physical footprint in, in existing data centers. And you know, CIOs and CTOs are being pressed to do more and more with, with less. Right. Ultimately, especially in the age of AI where they've got to figure out where to put that. You know, those Nvidia powered AI enabled systems, there's just no more room, no more space and they've got it. So they've got to figure out how to do more with less. And that's really a theme that we're seeing becoming quite common now across the set of clients that we work with.

‍

Patrick Moorhead: Marcel, has it changed a bit? Because, you know, I've been in tech for, I guess this is my 35th year, I'm old. Don't hold it against me. Some people call it experience, but you know, similar types of needs and requirements that I've heard. What makes RightNow different is it the giant power sucking sound of GPUs and generative AI that's making this even more difficult?

‍

Marcel Mitran: That's definitely a large part of it. I mean, I think everybody wants that additional competitive advantage of having AI Helping them grow and accelerate innovation business. The reality is, you know, there's a lot of existing footprint out there that's, you know, been around for a while and has sprawled over time and created significant complexity and debt. And you know, budgets aren't getting bigger, Data centers aren't getting bigger. Power grids aren't grading, and aren't getting more power. You know, this is an interesting statistic. You know, data centers worldwide are now sucking over 1,000 terawatt hours of power. There's just no more to give. Right. And so people are having to figure out how to deal with that. Right. While addressing the business needs of agility, resilience, intelligence. Right. So CIOs and CTOs are just really in a bad spot and trying to figure out how to deal with all of this.

‍

Patrick Moorhead: Yeah, Marcel, I've done more speaking with CIOs and CTOs this year than I did in the previous 15. And one of the biggest reasons is, listen, everybody wants to focus on outcomes and that will be true, but there's this technological layer that sits on top of it that seems to change every 90 days. And so I was in Europe, in fact last week I was in Italy and London. I did two CIO roundtables with, yeah, definitely in the Fortune 500. And they have expressed exactly what you had said related to efficiency. My follow up here is what are the main ways that you're seeing enterprises becoming more efficient?

‍

Marcel Mitran: Well, I mean, so if you sort of look at the landscape today around traditional distributed architectures, what you'll see is there's a ton of heavily underutilized infrastructure out there. I mean, what we're observing is, you know, average utilizations of like 16 to 22% on traditional scale out our infrastructure. And that, you know, there are very good reasons for why that infrastructure runs at those utilizations. I mean, you're typically, you know, potentially bound by memory. You're managing to peak utilizations on small granular pieces of infrastructure. So you can't aggregate workloads and get to a better sort of peak to average utilization as a result. And so if you think about that, that's like 84% of your cycles are spent essentially idling, doing nothing useful. Right. That's a huge, huge opportunity to improve on efficiency in the data center. And so that in my mind is where a lot of the low hanging fruit is today. We need to sort of take a step back and rethink this traditional sort of pizza box scale out design point for how we do data centers.

‍

Patrick Moorhead: Is it really planning for a peak where a lot of the gaps are, where you know, this is your peak and you know I'm going to, even though the average is down here.

‍

Marcel Mitran: Right.

Patrick Moorhead: I have to plan, I have to plan for peak, have to plan for peak.

‍

Marcel Mitran: And the peak to average is just a large, there's a huge difference between the two. Just because of the way, you know, we're, we're distributing workloads and giving every workload its own system, for instance. And that, that's definitely a big part of it. Yeah.

‍

Patrick Moorhead: So is part of the solution here being able to intersperse jobs that keeps your IT working all the time? Maybe some element of the batch, maybe some, you know, different types of functions in there?

‍

Marcel Mitran: Yeah, I mean that's the idea is if you can aggregate workloads so that you can really take advantage of sort of averaging out those peaks and valleys across a large set of workloads, that's where you have this magic of being able to reduce the peak to average to something that's much smaller. And that's largely the magic that sits behind Linux One. Linux One is really unique in its sort of scale out on scale up architecture. It's really sort of the way you want to think about it is it's like 85 metal servers that have this magic ability to share compute, IO and memory resources dynamically and non disruptively under the covers. So that as you sort of aggregate workloads across those 85 servers, you're able to really average out the peaks and valleys and really drive a much more efficient use of the infrastructure resources and on those systems. And that's where you know, we can get to 95% utilization on average with a Linux One system.

‍

Patrick Moorhead: Oh, I love that 85%. Hey, can you, for, for the audience out there that may not be as familiar with the latest Linux One out there, talk about what you mean by the 85 servers, because possibly some of the viewers back here, if you're, or if you're just listening in a podcast, might be thinking 85 servers, 80, 85 boxes, 85 trays that are going down. A fleet of servers here.

‍

Marcel Mitran: Yeah, a Linux One system is essentially a large uber Linux server that you can divide up the resources within the server into 85, what we call logical partitions. And those 85 logical partitions can maintain an air gap like isolation between them. But what's really neat about that is that you can actually share the compute, the IO and the memory underneath those 85 partitions and you can dynamically reassociate them non disruptively for the workloads that are running across those 85 partitions. So it is essentially like 85 metal servers in a box that have this magic ability to share all their resources and dynamically move the resources around.

‍

Patrick Moorhead: That makes a lot of sense. I love l pars. That's been, you know, it's, it's been something that I've been, you know, learning about forever. Very cool. Hey, let's dive into, I would call it a related, some would say tangential type of conversation that's about sustainability. Okay. And it's interesting. I always like to say it depends on what year we talk about sustainability. It's either lead with green is good for the world or if it's man, let's save money. And then, you know, we come back another year and it's hey, let's find a way to save money and oh, it's green.

‍

Marcel Mitran: Right.

‍

Patrick Moorhead: But you know, my trips to Europe, I mean it's, it's very much. Even sustainability is really high on that list. So how does sustainability fit into this efficiency conversation for a modern infrastructure?

‍

Marcel Mitran: I think good corporate citizenship continues to be a reality. Right. Every enterprise has a net zero strategy stated and that makes a ton of sense. We all want to do good for the world and so on. But I think what a lot of the enterprises are realizing is there's actual value from a bottom line perspective from that as well. So it's sort of like icing on the cake when hey, we can be good corporate citizens, but we're also helping ourselves become more agile.

‍

Patrick Moorhead: Right.

‍

Marcel Mitran: Enterprise because we're doing more with less ultimately. So yes, I think it's actually both. I definitely see that both are kind of playing out now in the industry.

‍

Patrick Moorhead: Yeah, I'm seeing that too. And it's funny, sometimes we get asked to predict energy prices, which is super tough when you see a quadrupling of the energy drawn from hyperscale data centers and us kind of juggling. It was crazy. In one week we went from in the United States, nuclear power bad to literally every major company. It's now on their green list. And I think what that did is really drove the conversation about how much power we're going to be generating here and need, need to do that. And you know, there's a lot of fear out there that, you know, the average citizen is going to have to pay more for their power. Like why should I pay 20 to 30% of my power to drive my home when I might not be getting benefit out of it. So I absolutely think that that's one conversation that's not getting a lot of play here. That absolutely, that absolutely will. So how does Linux One fit into that? I know you talked about, you know, first of all, workload utilization. That plays into that. If I can get to 85% utilization rate there as well. I don't want to put words in your mouth, but I would also guess that Linux One consolidation is an unlock too. But I'll let you answer the question here.

‍

Marcel Mitran: Oh yeah, no, I mean from a consolidation perspective it's a great story, right? I mean when you get to that higher utilization using an order of magnitude less cores, there's inherent savings there from a total cost of ownership. And if you look at the recent study that we published with the Signal 65, you know, there's. We saw 44 to 75% reduction in total cost of ownership. A large part of that actually is software. 94% reduction in software cost. And that gets down to just how software is licensed and priced, right. It's typically priced per core. So if you need 10 times less cores to do the same amount of work, that turns into a dramatic savings. From a software cost perspective. The added efficiency is also from that consolidation of energy. So we're seeing 65 to 91% reduction in energy required to run those workloads and an up to 82% reduction in data center space. So it's a savings on how better efficiency just translates into savings on all those dimensions, be it total cost of ownership. Yeah, Be it space, be it power. Right. Just doing more with less. Right, yeah.

‍

Patrick Moorhead: It also seems like, you know, a typical workflow might be okay, I've got Linux, I'm doing Linux One workloads. I need to ETL a lot of the data out to do AI on that. And you know, there's energy that goes into it. You've got, you know, not to mention the potential security challenge of moving, by the way. That's true. Moving data anywhere.

‍

Marcel Mitran: Anywhere.

‍

Patrick Moorhead: But it seems like there is an efficient play of running certain AI workloads on top of an IBM Linux one too.

‍

Marcel Mitran: Oh absolutely. I mean, absolutely. I think we just announced the availability of the SPIRE accelerators, right. Which are integrated AI accelerators for inferencing workloads. That just translates into saving on data movement, reducing risk around data movement of course as well, reducing latency. So that in itself also a great story for how we're improving efficiency around AI scenarios, especially where you're talking about data Serving and application serving scenarios.

‍

Patrick Moorhead: That makes sense. Yeah. I love the spire architecture. I mean first of all you've got on chip if you need kind of the maximum bandwidth tightly coupled to the CPU and then you've got the spire cards if you want a little extra oomph to be able to do more there. And it just makes sense to me, particularly across a certain set of workloads. So it's been really good to see Linux 1 a step up here. Arcel, this is great and this is a great, I wouldn't call it strategic or philosophical conversation here, but obviously you have clients that are reflecting all of this data and the thesis in real life. Can you talk about a few of them?

‍

Marcel Mitran: Absolutely. I mean we've got a long list of client references in this space. Pat1, for instance, is one of the largest casualty insurers worldwide out of Asia. They were an X86 only shop. Their regulators were telling them they need to address their carbon footprint. It was out of control. They transitioned to Linux one everything and they reduced their electricity consumption by 70%. Their software cost came down by 90% and their floor space came down by 75%. Another. Sorry, go ahead.

‍

Patrick Moorhead: No, I just said that's awesome. I, I love the X86, you know, getting off of X86 and nothing against X86, but that, that is where you can unlock a lot of, a lot of these efficiency gains.

‍

Marcel Mitran: Yeah. And so we've got countless examples. Like another one would be a European financial institution. They moved their Oracle database off of x86 to Linux 1. They saw 15 to 1 core consolidation, 70% reduction in CO2 and a 60% reduction in software licensing. So I could go on all day with examples like this. It's just, it's proven technology. We've seen it over and over again with clients.

‍

Patrick Moorhead: Marcel, are these, are these case studies up on your, up on the Linux One website?

‍

Marcel Mitran: They are, yes, they are.

‍

Patrick Moorhead: Awesome. I appreciate that. Yeah. So everybody out there, we are not talking about just theory. This is actual results. I loved, you know, Signal 65 results are one thing. I'm a big fan of Signal 65. I co founded that company. But also putting it in real life makes a huge difference. So with that I want to thank you for coming on the show here and yeah, keep, keep innovating out there with Linux 1 and whether it's, you know, efficiency or effectiveness or everything in between, you've been doing it for, for years and it's just, it's great to see you stepping up to the plate and doing even more.

‍

Marcel Mitran: Awesome. Thanks Pat. Thanks for having me.

‍

Patrick Moorhead: Thank you. So that's Marcel Mitran, IBM fellow and CTO IBM Linux One. We are talking about IT efficiency in the context of Linux One.  Hope you enjoyed this conversation. Check out all the content about IBM Linux One out there on the Six Five Media site, but also my analyst firm, Moor Insights and Strategy. This is something that analyst Matt Kimball talks about already. So thank you for tuning in and take care.

‍