Transcript
Truth in IT: Hi Mike. Matchett Small World Big Data. And we are here talking with IBM again today about some of the latest and greatest things they're rolling out with storage. We've got stuff on AI, we've got stuff on cyber resilience, we've got stuff on detecting corruption. Great topic. So stay tuned. Hold on. Hey, Scott, Welcome back. Scott Baker [IBM]: Hey, Mike. Thanks for having me here. Truth in IT: So great things are happening in IBM storage. You've got some stuff coming out that's talking about cyber resiliency and helping people out with that. We've got some things to talk about with AI operations and some stuff to talk about. You maybe even mentioned sustainability a little bit for some folks, but tell us tell us a little bit about what you think are the the hot trends going on in storage today? What what are people really looking for? Scott Baker [IBM]: Well, Mike, that's a loaded question. I know we have a whole lot of time to get through all of that. So, you know, we're still holding pretty true from an IBM perspective about being the best cyber resilient storage in the market today. And in fact, you know, just at VMware itself, when we went to explore in the United States, we interrupt introduced this idea of being able to do real time data corruption detection that applies to any kind of a workload, but specifically how that applied to virtual machine environments, because the last thing you really want is your virtual machine environment to piggyback those kinds of those kinds of threats as it moves via an automated vmotion or storage Vmotion activity from one server in your VMware environment to another. But at the end of the day, what we were looking to do here is as data is being written into the storage array itself, how do we expand our IOPs capability to use machine learning to really take track of how that data is being written and the behavior and the relationship between the host and the back end storage array. So that using entropy, we can determine anomalous, you know, behavioral changes that could be indicative of a cyber threat and alert the IT staff accordingly. So I think cyber resiliency continues to be one of the leading trends going into the market moving forward for storage. And then who doesn't love AI? In fact, when I found out that you and I were going to get a chance to talk, I jumped over to ChatGPT and I did not do that. Truth in IT: But how do you know I'm real? How do you know I'm real, Scott? I'm a deep fake today. Yeah. Scott Baker [IBM]: This is the strangest thing, Mike, is that the first thing it said was, if you're going to speak to Mike, make sure you wear a green shirt. So anyway. But, you know, for business tends to be a big focus right now within IBM and certainly in the market itself. And one of the things that I definitely believe that the storage industry, the overall infrastructure industry as a whole, supporting AI based workloads has to begin to think about is helping organizations operationalize AI, not just helping them store data or throwing consulting services at them to try to figure out how to, you know, institute an AI model, etcetera. But how do you truly operationalize it? Because until you can operationalize something, you can't monetize it. And I think that's going to be the next big trend in storage. I think it's going to drive a lot of innovation in terms of computational storage requirements, interoperability within an ecosystem that knows how to not only, you know, store the data, but also provide the services that help you to extract information or inject value added information into the data set itself so that it can be appropriately loaded into whatever the AI workflows are and be processed accordingly, leading into what I would believe is the most important, which is how do you provide appropriate governance and and overall management. Truth in IT: Yeah, I just want to stop you there and just talk about So just to recap a little bit, you started talking about corruption detection with AI ops, right? So you're putting some, some some AI goodness into the storage ray itself where you you said entropy, which I think is, you know, a measure of randomness. You're basically looking for stuff that's getting encrypted just from a data perspective, and you're going to tell people that like, Hey, did you mean for this to be happening or not? Which, which is great. So the storage rate starts to get smarter from AI, but now you've got these concerns where you're talking about say, how do I look at businesses and say, you guys are all out there trying to use our storage and trying to use our infrastructure to run workloads, and we're going to try to help you do that. And I know you mentioned a few things, but like what does that what does that look like in the future? If someone says, hey, you know, I've got I've got, you know, flash systems today, I've got I've got a scale, I've got some things. What what does it look like if I try to put workloads in there tomorrow and how are you going to help me do a better job with that? Scott Baker [IBM]: So, you know, when we think about when we think about workloads from the storage perspective, whether it's IE, high performance computing or traditional database architecture, the storage unit itself doesn't really see the workload as being any different. One thing I would say, though, is AI tends to be both performance intensive as well as capacity sensitive with respect to what it needs from the back end. Supporting architecture. From a storage perspective, I think what AI for business will do is really drive two areas of focus. One, having the broadest ecosystem that you can have in terms of interoperability with different hosts, the ability to hold on different kinds of application types or excuse me, data types and support different kinds of applications. I think the other thing that it does is it drives an equal importance on being able to deliver not only a dedicated appliance for your storage, traditional disks and controllers and, you know, power supplies and whatnot, but also the ability to provide for software defined. And when we say software defined, it also means a willingness to have available to the market a truly unified offering that's capable of deploying both block and file, but also object storage as well. And in this particular case, as we look at our own portfolio, Seth becomes a really great software defined unified storage offering that we can inject into the Watson Exadata offering as a licensable component. Truth in IT: Yeah. So all that, all that starts to come together and says, if I, if I'm an organization going, I need to build my center center practice or my my center of excellence and get some best practices going. I can I can find the infrastructure that I need from here and put that together even in a bigger sense, right? You're not just the storage, not just the power systems, but consulting and some other things I think you guys can offer. Scott Baker [IBM]: Yeah, that's right. So as we begin to think about that, that ecosystem that I had mentioned, beyond the consultative aspect, being able to talk to experts in the areas of generative AI or AI based workloads, you know, the ecosystem has to move up stack, you know, to the compute layer and to the networking layer as well. And then as we also begin to think about not just being able to integrate with them, but how do we consume them. Because I think what I will do in terms of infrastructure is it's going to demonstrate the absolute need for utility or elasticity, meaning I don't need a big GPU farm, but I definitely need it at times and I certainly don't need a whole lot of capacity, but I need it at times. And I think we'll also see a necessity for storage products to be thinking about data, you know, the transitory nature of data. So we're more semi persistent as it moves along that chain of use before it gets to the final outcome. Truth in IT: Yeah, right. So we're going to have more data, more of it's going to need to be governed, but we can't store it all permanently in a frontline system on primary data. So we're going to have to get smarter, our systems will have to get smarter and know where the data needs to go. I'm glad you mentioned some GPU GPU capabilities because those are key for a lot of folks too. They can't just buy more GPUs. Nvidia is not making enough of them. They have to make 100% use of the ones they have. So that becomes really key on there. And and guess, guess we started off with just a couple of minutes left here talking about cyber resiliency. So I'd like to circle back around to that. Uh, when, when you talk about, I mean, obviously data is important, but it all has to be secure. And no matter what data we're talking about, right? So what are we what, what should we be thinking about in terms of cyber resiliency today? Lots of people are talking about I've got to make immutable snapshots and I can do this and that. What's really what's really needed? What are people really looking for? Scott Baker [IBM]: You know, I definitely believe that there is a continued need to understand where they are most vulnerable. Right. What does my attack landscape actually look like? And, you know, as we mentioned previously, IBM offers a free cyber resilience assessment tool that helps organizations, you know, regardless of what storage you've chosen, regardless of what servers you're using, networking components you have. The whole goal for us is to make sure that we take a moment to sit down with you in a consultative setting and we work through to identify where those those different kinds of of threat entry points might be. And then we help you organize an approach, a roadmap, if you will, to shore those up. You know, should you be using something like a flash system, for example, we'll help you set up the appropriate defenses to safeguard that data, whether it's through immutable snapshots that are stored in a secured and isolated recovery environment. That's what I would consider to be more logically airgapped from the rest of the system to actually integrating it with your ITSM investment so that you begin to use correlation as you look at that inline corruption detection and then the ability to respond autonomously based on feedback that might be coming from your security software stack that's trying to manage the the data points that are coming in from the storage, from the compute, from the networking, etcetera, to getting you to a point where you can actually recover very quickly as you respond to these kinds of cyber threats. And in fact, on the 12th of September, we've announced our cyber recovery guarantee. And now let's come back around to your real question is this As I begin to think about AI based workloads, you know, what does this mean? Well, I think the notion of a threat or an intrusion to an AI workload isn't so much as to whether or not ransomware gets in. It's whether or not malicious data is used to infect the the corpus of information that's used to train the AI model. And I definitely believe that we will redefine maybe a maybe a new cyber category, maybe maybe a veracity breach where you're trying to inject malicious data to corrupt the AI model itself. And so the importance of being able to detect when you have data corruption coming in will be absolutely critical as a front line defense for protecting whatever the workload is that you're running. And in this case, we're talking about AI, but also being able to create the appropriate recovery checkpoints where that veracity has been validated along the way so that, you know, once you get that data back, it's of the highest referential value or quality possible so that it can be re-instituted into production. Truth in IT: Yeah. So it's not just about creating a snapshot of something we've got to have make sure we've got safeguards around the copies, we've got to have some verification, we've got to have some guarantees about getting recovery back out of those snapshots and we might have more concerns about somebody actually trying to put bad data into our systems. And I think, you know, you you mentioned in the case, I think we've already seen some of that with the self-driving cars and stuff, people trying to spoof those and say, hey, you know, there's a stop sign there or there's not a stop sign there. No, that's not a cliff face at all. Don't worry about that. Right. So definitely, definitely some scenarios coming that people haven't haven't yet had to protect themselves against. So it's great. So so you mentioned that there's a cyber recovery guarantee. Just a little bit about that. Scott Baker [IBM]: Yeah. So one of the things that we wanted to do is kind of put our our mouth where our marketing is, if you will, at least our commitment. We've had the safeguarded copy snapshots and market for a little over a year now, and they're performing better than we absolutely expected them to perform. Customers love them. And what we wanted to do was just say, Hey, look, we guarantee you that you'll be able to recover from a breach or a cyber attack with respect to one of these safeguarded copy snapshots within 60s. And that's the first phase of the cyber recovery guarantee. And that'll be maybe a teaser for you to come back and listen to the next update when we begin to take a look at that overall security framework and where we fit and how we begin to expand that guarantee out, you know, not just from the first point of recovery, but actually getting into other points of recovery that might be, you know, even farther down the line, which could include different kinds of media, different locations, things of that nature. Truth in IT: I mean, it's a really businesses want, right? They want cyber resilience. The particular bits and bytes of how the different medias are implemented are decisions at a lower level, right? That's right. Scott Baker [IBM]: And I would also say if you look at this from a, you know, from a business to human perspective, because at the end of the day, you may sell, you know, a business may sell to another business, but, you know, you're asking a human to put their career on the line, you know, to make the recommendation to go with your product. And I definitely think that the guarantee itself is a form of confidence building with the business, but with the individuals therein, that the investment that they're making is going to protect their own careers as well as their business in the event that an. Should occur and they need to respond quickly. While the business continues to operate. Truth in IT: All right. I mean, that's great. And I'm going to finish off here by asking you if someone wants some more information about this or follow up on either the things we're talking about or cyber resilience. What would you send them? Scott Baker [IBM]: Right. So for the guarantee itself, if you just hit ibm.com slash slash system, you can learn all about that. We'll also be pumping out into social media through blog post or LinkedIn banners, etcetera. The VMware Explorer event. And I'm sure you'll be able to do a replay from that. And that way you can see what was discussed on stage. Truth in IT: All right. Check it out, folks. Thank you, Scott, for being here today. Scott Baker [IBM]: You think you might? Truth in IT: Yep. Is always interesting. Come on back and talk to us some more about it. Definitely want to dive into a lot of the topics you just scratched the surface of today. Take care and stay tuned, guys.
