 Hi, everyone, welcome back to theCUBE's Palo Alto studios. You're watching the IBM storage summit live. We're so excited to be able to bring you this new format in our Palo Alto studios, and we're going to be talking about AI and a hybrid cloud environment. You know, years ago, IBM had an event called IBM Edge. Rob, theCUBE would go there. We went to, I think, every IBM Edge, certainly the first one and I think the last one. So up through COVID, COVID changed everything, but it was an opportunity to hear from the experts in the storage world, in the ecosystem, what the latest innovations were. I loved that was one of my favorite shows. This is like a mini IBM Edge, but so much has changed. We talked about hybrid cloud back then, but you know, it was different than it is today. It wasn't certainly as seamless. You didn't have that true cloud experience. We're going to talk more about that and we're going to talk about scale with Dave Wolford who's the worldwide senior product manager for IBM storage and for AI and cloud scale and John Zawis Towsky, who's a global system solution executive at Psycomp. Gents, welcome to theCUBE. Good to see you here in Palo Alto. Thank you. So good to be here. So John, what is Psycomp? Tell us about your company. Psycomp is a platinum IBM business partner. I've been in business for almost 30 years and we are a global company. So we're in 46 countries delivering hardware, software services to 150 countries across the globe. Cool, well, thanks for your time and thanks for being here. All right, Dave, so let's get into it. Sometimes people get a little confused the names, the names change. So tell us what's new in your world. Talk about scale. Okay, so what is scale? Well, scale has been changed over the years. It's a platform that's been around. Originally it released as GPFS and then renamed Spectrum Scale and now recently IBM Storage Scale. And the reason why that's important is because really the technology, the base technology hasn't changed, but really the focus and what we brought because it's constantly evolving, constantly being enhanced. What is storage scale? So it's really a high performance, parallel file and object system that has global data platform. That's really what it is. And it's a whole idea of connecting data. It's not just high performance but it's really this connectivity. And it's part of a global platform and a portfolio of solutions that we basically cover with their file and object which also includes our recently brought over IBM Storage Ceph. So I think I'm correct in an assertion that probably most of the world's data is stored in file format, probably. But object is like the hot growth area, right? It's simple, get put. So bringing these two worlds together is really important. And if most of the data is there, certainly most of the cloud data is an object, but if that's where the world data is, you're going to be bringing the AI to that data. Correct. So what's the thinking and strategy around the fusion, no pun intended, of AI and all that data at scale? Well, you know, fusion kind of brings things together also more from the OpenShift or the container platform. And it's actually built on similar technologies with IBM Storage Scale as well as IBM Storage Ceph. And so that's kind of what Pete talked about and he brought that in together with the containers. What Storage Scale and IBM Storage Ceph really bring is really the platform and bringing it together. So, and we offer basically the file and object bringing those two protocols together onto a single platform. So Rob, I mean, Ceph was like this little jewel inside of Red Hat that kind of didn't get as much attention as it should have. So when IBM made the move to put Ceph actually inside of IBM in the storage business, that was an interesting move to me and sort of unleashes a lot more innovation. Yeah, yeah, I think like you got at with GPFS and Ceph being the underpinnings underneath that, that gives you, I would say, more flexibility in how your consistency, where it runs, all of the different things because to your point, I think even though a lot of things are in object, there's still files. I mean, if you look at things like Parquet files that are on object storage or what have you, if you're using something like Apache Spark or something like that, is that where you're seeing and maybe from the customer aspect, where are you seeing this be deployed from? Is it a lot of people looking to understand how to bring all their data together under one kind of namespace? Well, yeah, that's the nice part about Storage Scale. It is a global namespace. It is the place where you can handle all the data that you need and the fact that a lot of data is stored in object. Object isn't necessarily where you're going to get the boost in the performance, right? You're going to need to hydrate a scale cluster in order to feed those GPUs and feed that AI workload and get that to where you want it to be. So where's the pain for customers? Or maybe the better question, but there's two sides of a coin. There's the pain and you want to alleviate that pain, but then there's opportunity. Maybe that's where AI comes in. You know, AI obviously can drive productivity as we've been talking about all day, but what are you seeing from the customer standpoint? Where is their pain and where is their hope? Well, the pain is still silos of data in a lot of places. And that's what this platform, the global data platform solves, right? It gives you the reach into all those disparate silos, whether you're using an NFS protocol or a native protocol to the application and you can go in there and actually see that data and know where it is, as opposed to now trying to figure out where did that data land in that silo? Is it in this one, is it in that one? The global data platform brings all that together and that's what's stored scale. Really, it shines in that light. And so how are people deploying it? What are their prerequisites that they need to think about, you know, what role are you playing in terms of helping them adopt and accelerate that adoption? So on-prem, we definitely work with IBM and we deploy the IBM's ESS, which is their high-performance storage system with scale on it. But we actually built a managed service in the cloud using IBM's storage scale, where we actually deploy it, we manage it, we help the clients understand how it needs to be tuned for their workloads and take care of soup to nuts, basically, so that they don't have to manage it. They can do their application and run their business and not have to worry about how do I tune it? I'm adding new workloads. What does it look like? So they just, they see what? You're essentially, your interface, your platform, or is it? We actually built the interface through the marketplace in each cloud. So they can just go in and click on it, or they can go in and click on it and you know, contact us and we'll get involved with them and work through the workloads and actually build the cluster, custom cluster the way they need it. Scale powers that at the back end. Okay, so what's the roadmap look like? How should we think about when you think, again, we're bringing in all this AI capability in, GPFS obviously was designed to handle a lot of data from the start, right? There's no, I don't think there are architectural limits. I'm sure there are, but we're not hitting them today. You know, I think the roadmaps here now, I mean, that's really the big part, right? I mean, we brought in the four data services with the access services, the caching services, the management services and the security services, right? There are things in the roadmap that are coming, actually things that have been talked about all through today, right? I mean, you know, everything from flash core modules to Kubernetes and containers, you know, but we have a lot of the things. So like last session, Pete talked about the data catalog. That's also part of storage scale. So the cybersecurity with cyber vault and safeguard of copy, those are also part of storage scale. Yeah, that was what I was going to say is, how does that tie to this theme of cyber resilience? And I think the easy button and being able to be part of a managed service provider type of deployment is huge. Right. Because as we were talking with Pete, people want the easy button, right? How does this help them, you know, not have as many knobs to turn and how they really meet their cyber resilience? Well, I think that's the whole thing about bringing this all into, that's why we call it the global data platform. You know, it's a term that we continually use, but we've really latched onto it because what it does is it brings everything together. I mean, we can tie together other vendor systems. We tie together tape, we tie together cloud and we bring that all together. And that's the whole aspect. It's not just our storage, it's our storage plus. And that's what it really does. So it fits in so well with things like NVIDIA, NVIDIA. It fits in so well with Watson X, with cloud packed for data, all the IBM technology, as well as the open source technologies. You know, as well as the cloud and the object data and file data. I've always thought this technology, you know, had an affinity obviously in super computer environments. I wonder if you could discuss, you know, some of the prominent, you know, show off spots where you guys have this. But more, I'm interested in, it's almost like the AI and the high performance world are smashing together and going mainstream. And then, because we've talked about this, you know, you're only going to have a few like mega large language models. They're going to be a lot of really small AIs embedded and running in different organizations. What's your point of view on that? Well, I think that's a great point because I think that's what we're seeing. We're seeing this, all this need of high performance computing, which was more used for research and more for AI for research, bringing now into the enterprise, which has some different requirements, but still now grasping on to this high thing of high performance computing. So because AI needs a lot of data, which is what HPC required, as well as they need high performance. But now they need other things, such as they want to lower the cost. They want to actually also bring in that security aspect. Security in a HPC or research environment, they were closed off. They didn't really need the security, but enterprises need it, right? The data now is critical. So that's what's different. It's not just scratch-based data, which was so used in high performance computing. You're right, these labs are like air gapped. Right. You had a comment? Just if you go down the cyber resiliency path and the safeguarded copy path, for where you've got scale on-prem, well now you take your safeguarded copy and you ship it up to the cloud. And when you get that ransomware attack, you've got to figure out where it came from, how to even combat it and get back. Whereas if your data's already up there with that safeguarded copy, I can bring my minimum viable company up in the cloud because I bring my applications up, my data's already there, I hydrate and I go. The MVC. I heard this term a couple of times today. I was going to say, yeah, MVC. How about environmentals, John? I mean, how much of an issue is that? I mean, there's a lot of talk about it. Sometimes it's seen as virtue signaling, but this bottom line impact in terms of power and cooling and customers, especially when you bring AI in, you talk about NVIDIA GPUs, not only are they the market hot, they touch them, they're hot. A lot of water cooled stuff going on. So how important is that to customers that you talk to? Oh, it's very important in having the hybrid world is important, right? Because you can't keep building data centers. At a certain point, there isn't enough footprint, there's enough power, but yet you've got the cloud vendors that have figured that part out for you. And when we had the supply chain issues, right, where you couldn't get servers for 12 months, but you could go to the cloud. So using that model and using the AI capabilities in the cloud and being able to hydrate a high performance storage system up there to generate all that data is really where we see it going. And we've got a large company that's helping healthcare with AI in the cloud, 100%. They've moved all that data in there and they're analyzing all the radiology pieces and things for those clients. Well, you're talking data silo, I wonder if we could double click on that. Healthcare, tons of silos in that industry, one of the worst, even trying to get to your own healthcare data is sometimes difficult. So can you talk a little bit more about that example or any others that you might have? Sure, so that one particularly is several healthcare groups try to share certain data that they can, right? So you have to anonymize it so that you don't expose the PII and stuff, right? But to get to cancer research and to get things done faster, they all have to share information. And this platform gives them that capability, right? So even with the access to it in the cloud and having them all be able to get to those pieces because it's the global data platform where, okay, somebody in Europe doing research wants to see something that's being done at Stanford. Well, they can now do that, as long as you don't break the regulations, right? And all that kind of stuff. No, I think that's the whole idea of the global clean room, data clean room type of aspect of it. And are you seeing that pop up in other industries as well or what are your customers asking about? Well, yeah, I mean, if you just take the chip industry, right? They need to do verification as they shrink geometries of the chip, they generate more data, which means they got to do more verification to see, can this chip even be built? Well, in order to do that, you have to have the high performance and be able to burst to the cloud and run those types of AI algorithms and stuff on it. Customers have a lot of choice, right? You've got pure play storage vendors, you've got guys that do end-to-end stuff, you've got the cloud, you know, you can put data in the cloud. Why should a customer choose IBM? I think we've kind of summarized in four key areas. One is, and it's all based upon our four data access services. And the first one is really our access services with our high performance. That's the first one because you got to have performance. That's just really a must. So with our recently announced, I just wrote a blog on it this week on our new 125 gigabytes per second that we've done because we're optimizing the software, right? When you optimize it. I haven't seen of hired a number faster for a single node. So that's the first one. And then it's our multi access with our multiple protocols, accessing the same amount of data. Then our caching services. That's unique. And that's what we've been kind of talking about with all our uniqueness of bringing storage together, whether it's cloud data on prem data or IBM or to tape, right? You know, you ask about, you know, you were mentioning about our supercomputers or some of our large systems. They have huge amounts of data, sometimes as much of an exabyte. They can't afford to keep that all online and they don't need it all online, but they need to access it. So we're one of the biggest leveragers of IBM tape. You know, besides if you're gonna put it in a cloud, that's one aspect, but the other thing is you might want to put it on tape, but you want to be able to access it right away. So I think every one of those customers use tape with IBM storage scale to actually kind of combine that. And it's not just backing it up. It's keeping it live and active. Yeah, I think that's the key is there's still a ton of tape out there. Even if they call it cloud, there's still a lot of tape behind that. There is. And the next one is our management services. And that's with our cataloging and all our optimization and our compression capabilities. And then the security, which we've talked about with cyber vaults and safeguarded copy as well as our reliability with our six nines of availability. So those are kind of the, we summarize it really to make sure it's understandable and easy to explain. Just an observation too, I wonder if you guys agree. I mean, a lot of this AI work is going to be done, you know, in the cloud, sure, but a lot of it's going to be done on-prem as well for a lot of reasons. I mean, the cost, not the least. I mean, you talk to some customers, they have literally millions of cores and you're probably not going to have that be cost effective in the cloud, maybe do some experimentation, maybe do some development work. Are you seeing that in the? Oh, absolutely, that's really the burst workload, right? Okay, I want to test this, I want to see how this is, but a lot of the industry, a lot of the big players have thousands of cores on-prem already that they're already using and generating. The nice part about it is, you know, if they've got to do a refresh in some of those cores, they can burst to the cloud and run that load and then move the data back down and keep massaging it and whatnot. I don't think anybody will ever be in the big side of the house where you've got the large enterprise clients will be 100% cloud, right? There's always the certain amount of data that they want to keep to themselves and certain IP that they just won't move up. Yeah, IP leakage is a big concern, obviously cost. And a lot of times that's where the data is. Right, right. Gentlemen, great to have you here in our live studio. Thanks so much for coming in and taking the time with us. Oh, thanks for having me. Thank you. Yeah, really welcome. Okay, up next, Sarbjit Johal is in the house. He's a well-known influencer and analyst and we're excited to have him on. Keep it right there. You're watching IBM's storage summit live on theCUBE from Palo Alto.