 Hi everyone, this is Dave Vellante. We're digging deeper into the world of database. You know, there are a lot of ways to skin a cat and different vendors take different approaches. And we're reaching out to the technologists to get their perspective on the major trends that they're seeing in the market. Because we want to understand the different ways in which you can solve problems. So look, if you have thoughts and the technical chops on this topic, I'd love to interview you, just ping me at Dave Vellante on Twitter. A lot of ways to get a hold of me. Anyway, we recently spoke with Andy Mendelsen, who is Oracle's EVP and he's responsible for database server technologies. And we talked a lot about Oracle's ADW, Autonomous Data Warehouse. And we looked at the cloud database strategy that Oracle is taking and the company's plans and how they're different, maybe from other solutions in the marketplace. But I wanted to dig deeper. And so today we have two members of Mendelsen's team on theCUBE. And we're going to probe a little bit. George Lumpkin is the Vice President of Autonomous Data Warehouse and Neil Mendelsen is the VP of Modern Data Warehouse, that business for Oracle. They're both 20 year veterans of Oracle. When I reached out to Steve Zavanic, who's a colleague of mine for many years, he told me how, he's always told me how great Oracle is relative to the competition. So I said, okay, come on theCUBE and talk about this, give me your best people. And he said, whatever these two don't know about cloud data warehouse, it isn't worth knowing anyway. So with that said gentlemen, welcome to theCUBE. Thanks so much for coming on. Thank you. Hey, glad to be here. So George, let's start with you. And maybe we could recap for some of the viewers who might not be familiar with the interview that I did with Andy, in your words, what exactly is an autonomous data warehouse? Is this cloud native? Is it Oracle buzzword? What is it? Well, I mean, autonomous data warehouse is Oracle's cloud data warehouse. It's a service that's built to allow business users to get more value from their data. That's what the cloud data warehouse market is. Autonomous data warehouse is absolutely cloud native. This is a huge misconception that people might have when they first sort of hear about something this service because they think this is a Oracle database, right? Oracle makes databases. This is the same Oracle database I knew from 10 years ago. And that's absolutely not true. We built a cloud native service for data warehousing, built it with cloud features. You know, if you're understanding of the cloud data warehouse market is based upon how you thought things look 10 years ago, like Snowflake wouldn't have even existed, right? You can't base your understanding of Oracle based on that. We have a modern service that's highly elastic, provides cloud capabilities like online patching, and it's fully autonomous. It's really built for the business users so they don't need to worry about administering their database. Yeah, so I want to come back and actually ask you some questions about that, but let me follow up and talk about sort of the evolution of the ADW. And where did you start? I think it was 2018 maybe? Where you came from, where you are today? Maybe you could take us through the technological progression and maybe the path you took to get here. Hey, so 2018 was when we released the service that made it generally available. But of course, we started much earlier than that. And this was started within my product management team and other organizations. We really sat down with a blank sheet of paper and we said, what should a data warehouse in the cloud look like? You know, let's put aside everything that Oracle does for its on-prem customers and think about how the cloud should be different. And the first thing that we said was, well, you know, if Oracle writes the database software and Oracle builds its own hardware and Oracle has created its own cloud, why do we need customers to manage a database? And that's where the idea of autonomous database came from that Oracle is managing the entire ecosystem. And therefore we built a database that we believe is far and away the simplest to use, simplest data warehouse in the market. And that's been our focus since we started with 2018. And that continues to be our focus, looking at more ways that we can make an autonomous data warehouse as simpler and easier for business users to get more value out of their data. Awesome. I have one more question. And actually, Neil, you might want to chime in this as well. So just from a technical perspective, you know, forget the marketing claims and all the BS. How do you compare ADW to the so-called born in the cloud data warehouses? You mentioned Snowflake. You know, Redshift, is Redshift born in the cloud? Well, it was par Excel, but Amazon's done some good work around Redshift. I think BigQuery is maybe probably a better example because it was, you know, like Snowflake started in the cloud. But how do you compare ADW to some of these other so-called born in the cloud data warehouses? I think part of this you mentioned, I mean, Redshift wasn't born in the cloud. It was, you know, a code base taken from a prior company that was an on-premise company. So they adapted it to the cloud, right? And, you know, we have done, as George said, much of the same, which is, you know, our starting point was not another company's code base, but our starting point was our own code base. But as George said, it's less about the starting point and it's more about where you envision the endpoint, right? Which is that, you know, whatever your starting point is, I think we have a fundamental different view of the endpoint. Amazon talks about how they're literally built for, you know, the cloud built for developers, right? Builders, right? And, you know, Oracle wasn't first in the infrastructure business. We entered through our applications. And all of a sudden, you know, we began taking on hundreds and thousands and hundreds of even more customers that were SaaS customers. Underneath was the database and all the infrastructure. One of the things that we took away from that was that we couldn't possibly hire enough people, DBAs to manage all of the infrastructure below our applications customers. So one of the things that influenced this is that, you know, customers expect SaaS applications to just take care of themselves, right? So we had to essentially modify the infrastructure to allow it to do so as well, right? And we're bringing that capability to those people who, you know, may or may not have an application, but their interest is, you know, more of this self-service agility type of aspect. So it seems to me, and George, you were sort of alluding to this before. I mean, you mentioned Snowflake a couple of times, and then Neil, something you just said, I'm going to pick up on is, you've been around for a long time. And, you know, when I talk to the Snowflake people, they know Oracle, they, a lot of them came from Oracle. They understand, I think, you can't just build Oracle overnight and build in the capabilities that Oracle has in their recovery. And you talk to customers and, you know, you are the gold standard of, you know, especially mission critical databases, so I get that. But Neil, you just sort of hint on it is it takes a lot of people and skill to run the database. And so that, that's the problem that you're saying you were attacking. Is that, am I getting that right? Right, so the people that you talked about who originally built Snowflake came from Oracle, but they came from Oracle more than a decade ago. So their context is over a decade old, right? In the meantime, we've been busy, you know, building autonomous and many other capabilities, right? Their view of Oracle is a view that was back more than 10 years ago, right? They're still adding capability. So a really good example of this illustration is, you know, Oracle, as you said, is the most capable system that's out there has been for many years. We've been focusing on how to simplify that and how do we use machine learning embedded within the system itself? Because according to the concept of autonomous is that inside is this machine learning system that's continually improving, right? That's the whole notion. We're in Snowflake's case, they're still adding functionality. Last year, they added masking, which, you know, functionality they didn't have. But when they added the capability, they added it without, you know, the ability for a business user to actually take advantage of. There's no capability for a business user to actually find the information that needs to be masked. And then after the information is found, you require a technical person to actually implement the masking. In Oracle's case, we've had masking and those capabilities for a long time. Our focus was to be able to provide a simple tool that a business user can use that doesn't need technical or security experience. Find the data that needs to be masked, PII data, and then hit a button and have it masked for you. So, you know, they're still, you know, without this notion of a strategy to move toward the system to heal itself and to manage itself, they're just going to continue, as they continue to add more capability, they will in turn add more complexity. What we're trying to do is take complexity out while others are adding it in. That's an interesting, that's an ironic twist. That's an ironic twist. The interesting way to look at it, and I want to make this about Snowflake, but I mean, hey, I like what they're doing. I like them, I know the management and they're growing like crazy. And the customers tell me, hey, this is really simple and it's simple by design. I mean, to your point over time, it's going to get more and more complex. And we were talking to Andy, I think it was Andy was saying, you know, they've got the different sizes. You've got a shape, you know, they call it T-shirt sizes. And I was like, okay, I got a small, I got a medium and a large, maybe that's okay. But you guys would say, we give more granular, you know, scaling, I guess is the point there, right? I mean, George, I don't know if you can comment on that. It's just, it's a different strategy. You got a company that was founded, well, I guess 2015 versus one that was founded in 1977. So you would think the latter has, you know, way more function than the former, but George, anything you'd add to this comment? Yeah, I mean, I'm always amazed that there are these database systems that are perceived as cloud native. And they do things like sell you database sizes by T-shirt sizes as you described. I mean, if you look at Snowflake, it's small, medium, large, extra large, two extra large, but they're all factors of two. You're getting a size of your database of two, four, eight, 16, 32, et cetera. Or if you look at AWS Redshift, you're buying your database by the nodes. You say, how many nodes do you want? And in both those cases, this is a cloud native. This is saying we have some hardware underneath our database and we need you, Mr. Customer, to tell us how many servers you want. That's not the way the cloud should work, right? And I think this is one of the things that we did with Autonomous Data Warehouse. We said, no, that's not how the rules should work. We still run our database on hardware. We still have nodes and servers, but we should tell the customer, how many CPUs you would like for your data warehouse? You want 16? Sounds good. You want 18? Yeah, we can give you 18. We're not selling these to you in bundles of eight or bundles of six or powers of two. We'll sell you what you need. That's what cloud elasticity should be. Not this idea that, oh, we are a database that should be managed by IT. IT already knows about servers and nodes. Therefore, it's okay if we tell people, your cloud data warehouse runs on nodes. With the Norco, as Neil said, we want them, the data warehouse should be used by the people who want to actually analyze their data. It should be used by the business users. Well, and so the other piece of cloud native that has become popular is this idea of separating compute from storage and being able to scale those two independent of each other, which is pretty important, right? Because you don't want to have to pay for a chunk of compute if you don't need the storage and vice versa. Maybe you could talk about that, how you solve that problem and to the extent that you solve that problem. Absolutely, we do separate compute from storage with our Thomas Data Warehouse. When you come in, you say, I need 10 CPUs for my data warehouse and I need two terabytes of storage. So those are two dependent decisions that you make so they're not tied together in any way. And you are exactly right, Dave. This is how things should work in the cloud. You should pay for what you need, pay for what you use, not be constrained by having fixed sets of storage you have to use for a given amount of CPU or vice versa. Okay, go ahead, Neil, please. Oh, just to add on to that, the other aspect that comes into play is that, so you're starting going as X, whatever that happens to be. Over time, that changes. And we all know that workloads vary throughout the day, throughout the month, throughout the year by various events that occur. Maybe a close of the business at the end of the quarter. It may be a holiday season for retailers and so forth. So it's not only the starting point, but how do you actually manage the growth? Scaling up and scaling down. In our case, we tried, as George said, we abstracted that completely for the customer. Basically said check a box, which says auto scale. So if the system is required more resources, we'll apply more resources. And we do so instantaneously without any downtime whatsoever. Because again, people think in terms of, these systems have now become business critical. So if they're business critical, you can't just shut down to expand. Imagine it during the holiday season as your business is ramping up. And then all of a sudden you have to scale, right? And your system either shuts down, reboots itself, right? Or it slows down to the point that it's a crawl and all your customers get frustrated. We don't do that. You click a button, auto scale. And we take care of it for you, smoothing out those lumps, right? Without any technical assistance. And again, if you look at Redshift, you look at all these various systems, they require technical assistance to be able to figure out not only your initial status, but how you scale out over time. Interesting. Okay, so all this said, a lot of companies are using Azure, AWS, Google for infrastructure. Why would these customers not just use their database? Why would they switch to Oracle or add ADW? Well, I think Neil will probably want to add something. I want to start by saying a huge number of our existing autonomous data warehouse customers today are customers of AWS and Azure. They are pulling data from AWS and Azure and bringing it into an Oracle Autonomous Data Warehouse. And we built features. I focused on Prog Management. We built features for that. And so it's perfectly viable and it's almost commonplace for the very largest enterprises to be doing that. But then coming to the question of why would they want to do it? I don't know, Neil, do you want to? Yeah, yeah. So one of the things that we really see emerge here is a data warehouse doesn't generate the transactions on itself, right? So the data has to come from somewhere, right? And you ask yourself, well, where does the data come from? Well, in a lot of cases that data is coming from applications and increasingly SaaS applications that the company has deployed. And those are, you know, HR applications, you know, CRM applications, you know, ERP applications and many vertical applications. In Oracle's case, what we've done is we say, okay, well, we have the application, this transactional thing. We have the infrastructure from the autonomous data warehouse. Why don't we just make it really, really easy? And if you're an Oracle applications customer that's already running on the Oracle Cloud, we will essentially provide you the ability to create a data warehouse from that information, right? With largely either with a product and service or a quick start kit. You don't start from scratch. You start from where you are. And in most cases, there are many cases that where you are has data very much as George mentioned before, telcos, banks, insurance companies, governments, all of the data that they want to analyze. A lot of that data, guess where it's coming from? It's coming from Oracle applications. So it makes sense to be able to have both the data that's generated and the data that's being analyzed close to the same place. Because at the end of the day, the payoff pitch for any form of analysis is not coming up with an insight. Oh, I realized XYZ, but it's rather putting the insight directly into production. And that's where when you have this stuff spread all over God's green earth, trying to go from insight into action can take months, if not years. The reason that a lot of customers are now turning to us is that they need to be much more agile and they need to be able to turn that insight into action immediately, without it being a science project. Okay, thank you for that. So let's take them off. Like one of the top things that customers can get from Oracle Autonomous Data Warehouse that they couldn't get from say a Snowflake or a Redshift or a BigQuery or a SQL server or something. I appreciate you guys willingness to talk about the competition. Let's take them off. One of the most important things that we should know about that they can't get elsewhere. So first, I mean, we already talked about a couple of what we think are really the major themes of Autonomous Data Warehouse. The service is autonomous. You don't need to worry about managing it. Anyone can manage the data warehouse. The service is elastic. You can buy and pay for what you use. You know, those are just what we think of as being the general characteristics of Autonomous Data Warehouse. But, you know, when you come to your question of, hey, what do we give that other vendors don't provide? And I think that one angle that Autonomous Data Warehouse does a really good job is, and Neil was just discussing this, it focuses on the business problems, right? We have years and years of experience with not just database security, but data secured, right? You know, every cloud vendor can say, oh, we encrypt all your data. We have these compliance certifications, all of these things. And what they're saying is, we are securing your database. We are securing your database infrastructure. An Oracle, of course, has to do those as well. But where we go further is we say, hey, no, no, no, no, no, no. We know what business users want. They want to secure their data. What kind of data am I storing? Do I have PII data? Can you detect whether there's PII data and tell me about it in case some user loaded something that I wouldn't aware of? What kind of privileges did I give my users? Can you make sure that those privileges are right? And can you tell me if users were given privileges that they're not using, maybe I need to take them away? These are the problems that Oracles tackled in security over the last 20 years. It's really more about the business problem. So, yeah, some other, oh, go ahead. I'm sorry, I got so many questions for you guys. We can go back to that, because it sounds like there's a long list. We have nowhere to go. I want to pick up with George on something you said about elasticity. Is it true, pay by the drink? Do you have a consumption pricing? I mean, can I dial it up and dial it down whenever I want? How does that work? Yes, I mean, not to be too many technical details, but you say I want 14 CPUs. That's what your database runs at. You can change that default number anytime you want online, right? You can say, oh, okay, I'm coming up on my quarter in. I'm going to raise my database 20 CPUs. We just do it on the fly. We just adjust inside. What about the other way? What about coming down? Can I go down to one? You go down, you go down, you can go down to one. And you're not going to charge me for 14 if I go down to one? No, if you set it down to one, you get charged for one, right? Okay, that's good. In the background, you know, we are also allowing levels of all the scaling. You say, if you say, hey, I want to get charged for 14, and Oracle, can you take care of all the scaling for me? So if a bunch of people jump on at 5 p.m. to run some queries, because the executive said, hey, I need a report by tomorrow morning, we'll take care of that for you. We'll let you go beyond 14 and only charge you for exactly what you use for those extra CPUs beyond 14. Okay, thank you. Go ahead, Neil. Maybe if we had Andy talked about this when he was on the show with you last week, right? And he talked about this concept of a converged database, but let me talk about it in the way that we see it from a business point of view, right? Business users are looking to ask a variety of questions, right? And those questions need to be able to relate to both the customer themselves, the relationship that the customer might have with others. And today we talked about like the social network and who are influencers within that and then where they actually conduct the business, which is really, in every case, it's on some form of increasingly on a mobile device. So in that case, you want to be able to ask questions, which is not only who should I focus on, but who are the key influencers within this community, right? That could influence others. And does that happen in a particular place and time? Let's say pre-COVID, it might happen at a coffee shop or somewhere else. We can answer all of those questions and more inside of the autonomous system without having to replicate the data out to one system that does graphics and another system that does spatial a third system that does this. It's like a business user, it's like, wait a minute, come on, you're trying to tell me that I need a separate system and replicate the data, just be able to understand location and answer in many cases, yes, you have to have separate, which a business person says, well, that's absurd, right? Can't I just do this all in one system? You can with Oracle. So, and I'm not trying to be the snarky journalist or analyst here, but I want to keep pushing on this issue. So here we are, it's 2021, it's April, we're like a third of the way through the year. And so far, nobody has come out and said, okay, we're going to deliver autonomous data warehouse just like Oracle. So I asked myself, well, why is Oracle doing this? You guys answered to reduce the labor cost. But I asked myself, is this how they're solving the problem of keeping relevant a database that spans five decades? And you guys said, no, no, this is cloud native, born in the cloud, started essentially with a new mindset. But is this a trend that others are going to follow? And if so, why haven't we seen it, this idea of a self-driving database? Is it, why is it right now unique to Oracle? What's really going on here? So I think there's, it's a really interesting thing that's happening. It's not visible outside of Oracle. It's very visible for those of us who work inside of the development organization. You know, if you look at Oracle, I can't tell you, but I mean, I think it's safe to presume Oracle has the largest database development organization on the planet, right? I mean, it was kind of the largest database or most used database coming into for the past few decades. And what's happened is we pivoted to building a cloud platform. We're not just building a database. We're taking all of these resources that we have with all these expertise is building database software. We're saying we now have to build the platform to run and manage the database software in the cloud. And it's a little bit like, you know, I think to make things, make people relate to it a little better. There was a really good quote from Elon Musk a couple of years ago, talking about Tesla. Like everyone looks at the car, right? Tesla, the car is really great. The hard part of this is building the factory. And that's analogy holds for Oracle. What we're building is the cloud factory. And what we have transitioned is our database development organization is now building as robust a cloud as possible so that when we increase the number of databases by 10X, we don't add 10X more cloud ops people to manage it. We are ramping up developer building features to automate the management of our cloud infrastructure. And with that automation, we get better availability, less errors, more security. We give benefits to our cloud data warehouse customers with it. And I think it's something really important to realize, right? We build database software. We build, you know, an engineered system built for databases is called Exadata. We build a cloud platform. And these are really equal tiers in what we are building and developing today in 2021 from Oracle Database Development Organization. Well, you mentioned Exadata. I wanted to shift gears here a little bit and talk about, we're seeing this hybrid cloud, you know, on-premises clouds, they're finally gaining some traction. I got to give props to Oracle's cloud of customers really the early to that game. I think it was the first, in my view anyway, true same-same vision. It took you guys a little while to get there, but it was the right vision. And I think I always say about Oracle. People don't understand this. Oracle invests in R&D. Your chairman is also the CTO. You guys are serious about technical investment. So that, you know, that's where innovation comes from. But, and we heard during your recent earnings call, we heard some positive comments on this. So what's your take on delivering autonomous data warehouse on-prem? And how do you compare, would say, Snowflake and AWS in that area, Snowflake, Frank Slubin, I have my record saying, we're not going to do the halfway house. Forget it, we are always going to be in the cloud. We're never going to do an on-prem installation. AWS will see to date. Yeah, I don't think you can get a redshift, for instance, in outposts, but maybe that'll come. But how do you see that emerging? What's your difference there? Maybe, Neil, you could talk about that. Yeah, so, you know, I think, you know, customers in a lot of regulated industries, I still have concerns about the public cloud. And I think that when you hear statements like, you know, we're never going to do, you know, on-prem. Well, autonomous cloud of customer is not, it's not a classic on-prem solution. What it is, it's a piece of our cloud delivered in your data set. It's still the cloud software. Oracle manages it, Oracle, you know, the system itself manages itself, and we take care of that responsibility so you don't have to. The difference is that we can make that available in a public cloud, as well as in a private cloud, right? And there are so many use cases, you know, that you can imagine from a regulatory point of view, or just from a comfort point of view, where customers are choosing, they want the ability to decide for themselves where to place this stuff, as compared to only having one option, right? And, you know, you look at a lot of what's happening in the emerging world, where, you know, there are a lot of places in the world that may not have, you know, really, really high speed internet connections to make, you know, a public cloud feasible. Well, in that case, whether you're talking about, you know, an oil rig, or you're talking about something else, right? We can put that capability where it needs to be close to the operation that you're talking about, irrespective of the deployment option. Well, let me just follow up on that, because I think it's interesting that, you know, Frank Slupin said that to me, oftentimes around AWS, I say, never say never, because they'll surprise you, right? And I learned that with Andy Jassy, but one of the things that seems difficult on-prem would be to separate that compute from storage because you have to actually physically move in resources. I think about Vertica Xeon mode. It's not quite the same-same. So, I mean, in that regard, maybe you're not the same-same, and maybe that dogma makes sense for some companies. For Oracle, obviously you've got a huge on-prem estate. Thoughts on that? So, you know, clearly, you know, so typically what we'll do is that we'll provide additional hardware beyond what the customer might expect, and that allows them to use the capabilities of expansion, right? We also have the ability to allow the customer to expand from their cloud of customer into the public cloud as well, of which we have a lot of those situations. So we can provide a level of elasticity even on-premises by over-provisioning the systems while not charging the customer until they use only based on what they consume, right? Combine together with the ability for us to augment their usage in the public cloud as well, right? Where others, again, are constrained, right? Because they only have a single option. Well, and you've got the capital resources to do that as well, which is not to be overlooked. Okay, I mean, I've blown our time here, but you guys are so awesome, I think. Thanks for, I appreciate the candor. So last question, and George, if you want to throw in a couple of those other tick boxes, you know, the differentiators, please feel free, but for both of you, if you can leave customers with the one key point or the top key points on how Oracle Autonomous Data Warehouse can really help them improve their business in the near term, what would they be? Maybe, George, you could start, and then, Neil, you bring us home. Yeah, I mean, I think that as I said before, our starting point with Autonomous Data Warehouse is how can we build a better customer experience in the cloud? And I think this continues throughout 2021. And I think that the big theme here is the business user should be able to get value directly from their data warehouses. We talked a few times about how a line of business user should be able to manage their own data, should be able to load their own data warehouse, should be able to start to work with their own data, should be able to run machine learning model, build machine learning models, get that data. And all of that built in and delivered in Autonomous Data Warehouse. And we think that this is, you know, we see our customers, organizations large and small, the light bulbs starting to go on and how easy this service is to use and how complete it is for helping business users get value from their data. And just adding on to what George said, you know, the development organization has done a tremendous job of really simplifying this whole operation. But we also tried to do that on the business side. You know, when a customer has an on-prem situation, they're looking at moving to the cloud, whether lift and shift or modernize, they're looking at cost, they're looking at risk, and they're looking at time. So one of the things we look at is, how do we mitigate that? How do we mitigate the cost, the risk and the time? Well, this week, I think we announced our new cloud lift program. And the cloud lift program is what Oracle will provide through its cloud engineering resources around the world is that we will do, we will take the cost, the risk and the time out of the equation and Oracle will work directly with the customer or the customer's partner of choice, maybe an Accenture or Deloitte, and we will move them, right, you know, at little or no cost. In most cases, there's no cost whatsoever, right? We mitigate the risk because we're taking the risk on and we built a lot of automated tools to make that go very quickly, right, and securely. And then finally, we do it in a very, very short amount of time as compared to what you would need to do with, you know, because there is no redshift on premises, there is no snowflake on premises. You have to convert from what you already have to that, right, and but the company beyond the technological barriers that George talked about, we're also trying to smooth the operation so that a business itself can make a decision that not only do they not need the technical people to operate it, they won't need an entire consulting contract worth millions of dollars in order to actually do the movement to the cloud, right? Well, guys, I really appreciate you coming on the program and again, your candor to speak openly about, you know, your approach, the competitors. And so it's great having you. Really, really, thank you for your time. Appreciate it. And thank you for watching everybody. Look, if you guys want to come back, go toe-to-toe with these guys, say the word, you're always welcome to come on theCUBE. One thanks for sure, Oracle Sirius, when it comes to database. Thank you for watching. This is Dave Vellante. We'll see you next time.