 The history of Exadata in the platform is really unique and from my vantage point it started earlier this century as a skunk works inside of Oracle called Project Sage back when grid computing was the next big thing. Oracle saw that betting on standard hardware would put it on an industry curve that would rapidly evolve. And I remember the Oracle HP database machine which was announced at Oracle OpenWorld almost 15 years ago and then Exadata kept evolving. After the Sun acquisition it became a platform that had tightly integrated hardware and software. And today Exadata, it keeps evolving almost like a chameleon to address more workloads and reach new performance levels. Last April for example, Oracle announced the availability of Exadata X9M in OCI, Oracle Cloud Infrastructure and introduced the ability to run the autonomous database service or the Exadata database service. You know, Oracle often talks about they call it stock exchange performance level kind of no description needed and sort of related capabilities. The company as we know is fond of putting out benchmarks and comparisons with previous generations of product and sometimes competitive products that underscore the progress that's being made with Exadata such as 87% more IOPS with metrics for latency measured in microseconds, mics instead of milliseconds and many other numbers that are industry leading and compelling especially for mission critical workloads. One thing that hasn't been as well publicized as the Exadata on OCI is using AMD's EPYC processors in the database service. EPYC is not Eastern Pacific Yacht Club for all you sailing buffs, rather it stands for Extreme Performance Yield Computing, the enterprise grade version of AMD's Zen architecture which has been a linchpin of AMD's success in terms of penetrating enterprise markets and to focus on the innovations that AMD and Oracle are bringing to market, we have with us today Juan Loiza who's Executive Vice President of Mission Critical Technologies at Oracle and Mark Papermaster who's the CTO and EVP of technology and engineering at AMD. Juan, welcome back to the show. Mark, great to have you on theCUBE in your first appearance. Thanks for coming on. Yep, happy to be here. Thank you. All right Juan, let's start with you. You've been on theCUBE a number of times as I said and you've talked about how Exadata is a top platform for Oracle database. We've covered that extensively. What's different and unique from your point of view about Exadata Cloud Infrastructure X9M on OCI? Yeah, so as you know, Exadata, it's designed top down to be the best possible platform for database and it has a lot of unique capabilities like we make extensive use of RDMA, smart storage. We take advantage of everything we can in the leading hardware platforms and X9M is our next generation platform and it does exactly that. We're always wanting to get all the best that we can from the available hardware that our partners like AMD produce. And so that's what X9M is, it's faster, more capacity, lower latency, more IOs, pushing the limits of the hardware technology. So we don't want to be the limit the software, database software should not be the limit. It should be the actual physical limits of the hardware and that's what X9M is all about. Why won AMD chips in X9M? Yeah, so we're introducing AMD chips. We think they provide outstanding performance, both for OLTP and for analytic workloads. And it's really that simple. We just think the performance is outstanding in the product. Yeah, Mark, your career is quite amazing. I've been around long enough to remember the transition to CMOS from a mid or coupled logic in the mainframe era back when you were at IBM. That was an epic technology call at the time. I was, of course, steeped as an analyst at IDC in the PC era and like many witnessed the tectonic shift that Apple's iPod and iPhone caused. And the timing of you joining AMD is quite important in my view because it coincided with the year that PC volumes peaked and marked the beginning of what I call the stagflation period for X86. I could riff on history for hours, but let's focus on the Oracle relationship. Mark, what are the relevant capabilities and key specs of the AMD chips that are used in Exadata X9M on Oracle's cloud? Well, thanks. And it's really the basis of, I think, the great partnership that we have with Oracle on Exadata X9M and that is that the AMD technology uses our third generation of Zen processor. Zen was architected to really bring high performance back to X86, a very, very strong roadmap that we've executed on schedule to our commitments. And this third generation does all of that. It uses a seven nanometer CPU that is a core that was designed to really bring throughput, bring really high efficiency to computing and just deliver raw capabilities. And so for Exadata X9M, it's really leveraging all of that. It's implemented in up to 64 cores per socket. It's got really anywhere from 128 to 168 PCIe, Gen4 IO connectivity. So you can really attach all of the necessary infrastructure and storage that's needed for Exadata performance and also memory. You have to feed the beast for those analytics and for the OLTP that Juan was talking about. And so it does have eight lanes of memory for high performance DDR4. So it's really as a balanced processor and it's implemented in a way to really optimize high performance. That is our whole focus of AMD. It's where we've reset the company focus on years ago. And again, great to see the super smart database team at Oracle really a partner with us, understand those capabilities. And it's been just great to partner with them to enable Oracle to really leverage the capabilities of the Zen processor. Yeah, it's been a pretty amazing 10 or 11 years for both companies. Mark, how specifically are you working with Oracle at the engineering and product level? And what does that mean for your joint customers in terms of what they can expect from the collaboration? Well, here's where the collaboration really comes to play. You think about a processor and I'll say, when Juan's team first looked at it, there's general benchmarks and the benchmarks are impressive, but they're general benchmarks. And they showed, I'll say the base processing capability. But the partnership comes to bear when it means optimizing for the workloads that Exadata X9M is really delivering to the end customers. And that's where we dive down and as we learn from the Oracle team, we learn to understand where bottlenecks could be, where is there tuning that we could, in fact, really boost the performance above. I'll say that baseline that you get in the generic benchmarks. And that's what the teams have done. So for instance, you look at optimizing latency to our DMA, you look at just throughput, optimizing throughput on OLTP and database processing. When you go through the workloads and you take the traces and you break it down and you find the areas that are bottlenecking and then you can adjust. We have thousands of parameters that can be adjusted for a given workload. And that's again, that's the beauty of the partnership. So we have the expertise on the CPU engineering, Oracle Exadata team knows innately what the customers need to get the most out of their platform. And when the teams came together, we actually achieved anywhere from 20% to 50% gains on specific workloads. It was really exciting to see. So, okay, so I want to follow up on that. Is that different from the competition? How are you driving customer value? You mentioned some percentage improvements. Are you measuring primarily with latency? How do you look at that? Well, you know, we are differentiated but in the number of factors, we bring a higher core density. We bring the highest core density, certainly in x86. And moreover, what we've led the industry is how to scale those cores. We have a very high performance fabric that connects those together. So as a customer needs more cores, again, we scale anywhere from eight to 64 cores. But what the trick is that as you add more cores, you want the scale to be as close to linear as possible. And so that's a differentiation we have. And we enable that, again, with that balanced computer of CPU, IO, and memory that we design. But the key is, you know, we pride ourselves at AMD of being able to partner in a very deep fashion with our customers. We listen very well. I think that's what we've had the opportunity to do with Juan and his team. We appreciate that. And that is how we got the kind of performance benefits that I described earlier. It's working together almost like one team in bringing that best possible capability to the end customers. Great, thank you for that. Juan, I want to come back to you. Can both the Exadata database service and the Autonomous database service, can they take advantage of Exadata Cloud X9M capabilities that are in that platform? Yeah, absolutely. Autonomous is basically our self-driving version of the Oracle database. But fundamentally, it is the same database core. So both of them will take advantage of the tremendous performance that we're getting now. When Mark takes about 64 cores, that's per chip. We have two chips, it's a two-socket server. So it's a 128-way processor. And then from our point of view, there's two threads. So from the database point of view, it's a 256-way processor. And so there's a lot of raw performance there. And we've done a lot of work with the AMD team to make sure that we deliver that to our customers for all the different kinds of workload, including OLTP analytics, but also including for our autonomous database. So yes, absolutely. All that takes advantage of it. Now, Juan, I can't let you go without asking you about the competition. I've written extensively about the big four, hyperscale clouds, specifically AWS, Azure, Google, and Alibaba. And I know that, don't hate me. Sometimes it angers some of my friends at Oracle. IBM, too, that I don't include you in that list. But I see Oracle specifically as different and really the cloud for the most demanding applications and top performance databases and not the commodity cloud. Which, of course, that angers all my friends at those four companies. So I'm taking everybody off. So how does Exadata Cloud Infrastructure, X9M, compare to the likes of AWS, Azure, Google, and other database cloud services in terms of OLTP and analytics, value, performance, cost, however you want to frame it? Yeah, so our architecture is fundamentally different. We've architected our database for the scale out environment. So for example, we've moved intelligence in the storage. We've put remote direct memory access. We've put persistent memory into our products. So we've done a lot of architectural changes that they haven't. And you're starting to see a little bit of that. Like if you look at some of the things that Amazon and Google are doing, they're starting to realize that, hey, if you're going to achieve good results, you really need to push some database processing into the storage. So they're taking baby steps toward that, roughly 15 years after we've had a product. And again, at some point, they're going to realize you really need RDMA. You really need more direct access to those capabilities. So they're slowly getting there. We're well ahead. And the way this is delivered is better availability, better performance, lower latency, higher IOPS. So and this is why our customers love our product. And if you look at the global Fortune 100, over 90% of them are running Exadata today. And even in our cloud, over 60% of the global 100 are running Exadata in the Oracle Cloud because of all the differentiated benefits that they get from the product. So yeah, we're well ahead in the database space. Mark, last question for you is, how do you see this relationship evolving in the future? Can you share a little roadmap for the audience? You bet. Well, first off, given the deep partnership that we've had on Exadata X9M, it's really allowed us to inform our future design. So in our current third generation Epic EPYC is that, is really what we call our Epic server offerings. And it's a 7,003 third gen and Exadata X9M. So what about fourth gen? Well, fourth gen is well underway and ready for the future, but it incorporates learning that we've done in partnership with Oracle. It's gonna have even more through capabilities. It's gonna have expanded memory capabilities because there's a CXL Connect Express link that'll expand even more memory opportunities and I could go on. So that's the beauty of a deep partnership as it enables us to really take that learning going forward. It pays forward and we're very excited to fold all of that into our future generations and provide even better capabilities to one in his team moving forward. Yeah, you guys have been obviously very forthcoming. You have to be with Zen and Epic. Juan, anything you'd like to add as closing comments? Yeah, I would say that in the processor market, there's been a real acceleration in innovation in the last few years. There was a big move 10, 15 years ago when multi-core processors came out and then we were on that for a while and then things started stagnating, but in the last two or three years and AMD has been leading this, there's been a dramatic acceleration in innovation in this space. So it's very exciting to be part of this and customers are getting a big benefit from this. All right, Chance, hey, thanks for coming back in theCUBE today, really appreciate your time. Thanks, glad to be here. All right, thank you for watching this exclusive CUBE Conversation. This is Dave Vellante from theCUBE and we'll see you next time.