 Ladies and gentlemen, please welcome Oracle's Executive Chairman of the Board and Chief Technology Officer, Larry Ellison. Is this working? Is this working? It is working. OK. Hi, we're here to introduce the fifth generation of our engineered systems. Again, the idea is taking hardware and software and during the engineering phase, actually design them to work together. We do the integration, so you don't have to. But more than that, we locate functions at the right place in the stack. If it's a storage function, we locate it closer to the storage. If it's a compute function, it's close to the compute surface. And there's security through all layers, through the storage, and through the networking, and through compute. OK, this is our first generation, to be our fifth generation of engineered systems, called X5. And I'm pressing the button, and the slides aren't going forward. OK. All right, so let's look at data center trends. The biggest data center trend are these two socket Intel servers, which have a very, very low purchase price. People are building the core of their data center around these two socket Intel servers. And almost always running Linux. Yeah, some of them run Windows to run the Microsoft applications. But the bulk of the two socket servers, whether it's at a big enterprise account or at Facebook, run Linux, not Windows. So it's a lot of two socket servers, pretty much the cheapest server you can buy, running an open source operating system. And people are building that as the core of your data center. And it's very attractive because it's got a very low purchase price. You can make the argument. We make the argument that these do-it-yourself data centers are expensive. You need a lot of what you do the integration rather than we doing the integration. Engineered systems, we integrate all the layers, storage, networking, and compute together so you don't have to. However, we've really never aimed at the lowest possible purchase price with our engineered systems. We've never really competed for the data center core, those two socket Intel servers. We focused on something very different. We said, OK, we're going to do all the integration for you. And by doing that, by engineering every piece, the storage, then networking, the compute servers, and all the software in those three layers, we can deliver extreme performance. And our very first engineered system, Exadata, did just that. Delivered extreme performance. And we just have made it faster and faster and faster. As we went from disk storage to disk storage fronted by Flash to, you'll see with Generation 5, some options that are all Flash. There's just no disk at all. Or you can mix disk and Flash. But huge numbers of software optimizations from the storage server through the network, through the compute server to make it unnecessary for you to do the integration to deliver much better ease of use, much more reliability, much more security. And because they ran so fast, better cost performance, not lower purchase price, but if you consolidated a lot of your load on one of our engineered systems, we would argue, we would still argue, that you get a better cost performance that way, than you do with your data center core with a low purchase price. Okay, that's kind of the history of what's going on. Well, we introduced engineered systems. We didn't invent the idea. Teradata had a converged system. I think we call it an engineered systems. The industry calls it now converged systems. The idea of taking storage and networking and compute and putting them together. We didn't invent that idea. Teradata had its converged system or its engineered systems before we did. And then we came along with Exadata. And I think we did a better job. We certainly sold more systems than they did. And we still sell more systems than they do. And then we came out with other engineered systems, Exologic, Exolytics, more and more and more, all aiming at the highest performance and the best cost performance. There were even imitators. IBM came out with Pure Systems, which was very interesting. Didn't do very well, but interesting. I mean, copying you is the highest form of flattery. Then something very interesting happened in the industry is that Cisco systems developed a partnership with EMC. And they built, if you will, an engineered systems or a converged system taking storage from EMC and software from EMC, specifically VMware, Red Hats operating system and Cisco switches and then Cisco bought some two socket servers and put them together. And they had their integrated system, their converged system, their engineered system. But the thing that Cisco did, and they have been very successful, the big thing that Cisco did, was they went and decided to compete for that data center core with a very low purchase price. Interesting. I mean, they aren't close to us in performance, but they delivered a very low purchase price. And if you look at the two companies that are growing, or the two lines of compute that are growing today, Cisco UCS is growing, I think over the last four quarters, about in the mid 30 percentile. And engineered systems, Oracle's engineered systems is growing rather rapidly. Next slide, please. I can press the button, okay. So, I got it used to it. Two hands, I can do it. Okay, as I said earlier, our strategy for engineered systems in the very beginning was to deliver extreme performance, the highest performance. And once we did that, we would have better cost performance than anybody. You take a hundred applications, put it on exit data, and that's better than running those hundred applications anyplace else in terms of your overall cost, your overall cost of ownership, and the lowest purchase price. Moving on, you can just do so much on a single engineered system. But the purchase price of engineered systems was still an impediment, I think for some people, buying exit data. They believed us on the performance, but is it really true that, I mean, it's an expensive computer. It costs hundreds of thousands of dollars. A really big one can cost more than a million dollars. How could that be the cheapest one? Well, it is. But we're tired of having this argument. So we have a new strategy. We're gonna compete for that two socket data center core business. And from what I can tell the way you compete because it's two socket servers running Linux is you just have the lowest price in the industry. So our strategy, and you'll see it here, is to deliver not only the highest performance appliances and machines in the industry by a large margin, but now with the fifth version of our engineered systems to deliver by far the lowest price for the data center core. And that's what we're gonna do. And in that sense, we didn't invent that idea. I think Cisco and EMC did, but just like Teradata invented the database machine, we think we can do a better job. Okay, over here at my far left is the Oracle Virtual Compute Appliance, or VCA for short. And it is converged into infrastructure. It includes compute servers and networking. Actually, very advanced networking. And you'll see it's software-defined networking with very sophisticated hardware underneath it. But the real magic is unbelievably high performance and the lowest purchase price. So we've got compute servers, networking, and storage servers all in the same box. And, well, that's gonna be interesting. Maybe I do need my glasses. Okay, every year it gets blurrier and blurrier. Okay, so with our Virtual Compute Appliance, again, I said it combines hardware and software engineered to work together. Compute, networking, and storage. And it's highly available. It's fully redundant. It's fault tolerant. There's no single point of failure in this compute system. We do the integration, so you don't have to. So here's a combination of the engineered systems idea, appliance idea, that we do all of the integration. And the thing that's so attractive to purchasers. The other, and everything else is gravy. So we'll tell you that we can deploy applications in a couple of minutes where it takes other people because we have application templates that's part of our software stack. And other people take hours or days to deploy the applications. But that doesn't have to be how you make your decision because we've got the lowest purchase price in the industry. This is just gravy. This is just additional benefits. Reliability, security, performance. Those are all on top of the lowest purchase price in the industry. Okay, I think it's easier for me to read it here. Okay, here are the purchase prices. If you look at this VCA box, it includes 27, two socket servers, the latest Intel servers. That's almost a thousand cores. I mean, that's directly comparable to Cisco UCS. We've got the fabric internet to connect to connect to all of this storage with software defined networking. And includes the rack, it includes everything. Power supplies, but it's the whole McGillie, as my mom would say. And it's around half the price of Cisco looking at just the hardware component. Again, I'm saying this is just the hardware. This is not including software. Well, let's take a look at the software that goes with UCS. So you've got Red Hat Server, which has no purchase price, but it has an annual support fee. And you've got VMware, which does have a purchase price and a support fee. And then you have some management software. So to buy it, 265,000 annually, another 166,000 compared with our software price. So the hardware, much lower price. The software, much lower price. But of course you can't run this without storage. I'm pressing the button. Okay, I'm gonna throw this thing. Okay, next slide, please. I'm sorry, back of the, okay, my mistake. Okay, storage. You can attach, you know, VCA comes with different storage options. It has a fiber channel interconnect, so you're gonna connect to your existing storage if you want to. But we also will bundle in storage. And then you have some options. You can have network attached storage, or you can have a storage area network. You can have NAS or SAN. This is a comparison. This is the lowest cost comparison we come up with. And this is our NAS product. And it's up there for $46,000 plus 6K a year. And that's how, and again, right around half the price of what you get with UCS. So if you look at these totals, it costs, you know, for VCA plus NAS, plus all the networking, plus all the software. It's, but by the way, this is 27 two socket servers. So this is a lot of compute power, lot of storage, powerful network interconnect. It is about $600,000 with $70,000 support compared with well over, you know, more than double the price and almost triple the support costs. And by the way, the most interesting thing is the price comparison that I'm making is our list price to their discounted price. Not, how about that? So our list price is less than half of their discounted price. And by the way, to customers, we will negotiate and we will discount off the list price. This is a big secret. Don't tell anybody, but we actually will. So we are going to aggressively compete with version five of our engineered systems. We're going to aggressively compete for that two socket core of your data center. And the way you do that is you get very aggressive with pricing. And we think we've done that. Again, our list price compared with their discounted price. So we think this is, you know, this is a new strategy for Oracle and using virtual compute appliance integrated with software defined networking. And I'll go into that a little more detail and a couple of storage options to go after the data center, the data center core. Okay. First example I gave you was VCA connected to ZFS. Our network attached storage. We also have a relatively new sand product called the FS1 and this is the FS1 over here. And the FS1 combines our new strategy for engineered systems. Our old strategy was highest performance, best cost performance. Our new strategy is highest performance, lowest list price, best cost performance. We're just trying to make the decisions easier. So here is FS1 compared with EMC. And by the way, I'll show you that the FS1, we're not cheating, we're not picking out an expensive EMC box compete to compete with our low end FS1. This is our fastest, most expensive FS1 that you can buy. We don't make a flash system more expensive than the FS1. This is as much as you can spend. This is as fast as you can go. It is much faster than Cisco's fastest, excuse me, EMC's fastest flash system. And you can see apples to apples comparison comparing the same amount of capacity. We are less than a third of their purchase price. By the way, it runs much faster. Runs much faster. But that's gravy. Purchase price is one third of what EMC charges. And you got, well it's fully fall tolerant, no single point of failure, petabytes, petabytes of flash memory. And if you balance reads and writes and everyone knows that writes are more expensive than reads, but if you do 50% reads and 50% writes, it's still a couple of million IOs per second. And oh by the way, it has 10 times database compression, 10x database compression, which EMC does not have. So you can cut your capital expenditures by half by combining up virtual compute appliance and FS1 and you can deploy your applications in minutes, save a lot of time and a lot of labor and it runs anything in your data center. It doesn't, I mean a lot of people think our appliances are designed to just run the Oracle database or Oracle middleware or Oracle analytics. This runs everything. It runs your Microsoft applications, it runs anything. Anything in your data center. Okay, so the other interesting component of this, I've mentioned combining up VCA with FS1, you get very inexpensive, very powerful compute, very inexpensive, very fast storage. And they're connected by a very interesting network. Internally, and I really don't know if that slide's coming up next. Internally, this thing is interconnected by InfiniBand. Now you never see InfiniBand. In other words, when you talk to VCA, you talk to it via Ethernet. You never see InfiniBand. If you wanna connect a VCA to an existing EMC system or NetApp system, you do it via Fiber channel. All that capability is built in. But if you pair it up with FS1 or you pair it up with ZFS, and if you buy more than one node, and they, because the nodes have to talk to each other, the nodes are interconnected, all of those interconnections, node and compute node to compute node, compute node to storage. All of those interconnections are through a software-defined interconnect where the hardware, the actual physical interconnect is InfiniBand. So InfiniBand is interesting because it's much faster than any other interconnect, but it's also much more reliable than any other interconnect. And has guaranteed message delivery in a certain period of time, makes recovery much simpler. There's a whole bunch of things about InfiniBand beyond it just being fast. You know, it's, I don't know, 20, 30 years newer than Ethernet. 20, 30 years newer than ViberChannel. Ethernet and ViberChannel, I know they keep getting better, but these are very old technologies. The latest interconnect technology is in terms of the hardware connection, is InfiniBand. But on top of that, you wire this thing once, and then you use software. If you want to reconfigure your network, add more nodes, add more storage nodes, add more compute nodes, you reconfigure and expand by changing parameters in a file. The whole network is defined by software. So you wire it once and you're done. All that's for free. The software of DivineNet, it's all for free because we have the lowest purchase price. Okay, so if you look at what we're trying to do versus what Cisco has done very effectively, we're engineering all of our components so they work together. And we think the fact that we've engineered all of these pieces ourselves, one vendor, we designed them to work together, we built them to work together, we support them together, we update all the software, we test all the software to work together, by the way, with the hardware configuration. And then when we issue patches, it's all that's been tested. That's very hard to do when you have multiple hardware vendors and multiple software vendors. Cisco has its own software and it switches, EMC has its own software and it's disc controllers, VMware has its own software and it's on its own release schedule, Red Hat has its own software and its own release schedule. They're all issuing different patch releases at different times. The one responsible to making sure all those patches work together would be you. You just have to make sure you test before you actually install all these patches. The difference is with us is we actually will test all these pieces together before we ship them to you and we know that they work. Okay. We're about half the price of right now the most popular converged system in the data center core, which is Cisco UCS. And we're very interested in pursuing that business. And again, when you pair up VCA with FS1, you pay half as much, but you have to be willing to go twice as fast. Okay. We covered VCA and FS1. There are a lot of engineered systems I'm not gonna take nearly as long on each one of these. There's something called the Oracle Database Appliance. We sold thousands of them, not millions. We'd like to sell millions, but we've only sold thousands. To a variety of industries, ISVs love them. People develop their own applications, love them. They package their applications and ship them out on the Oracle Database Appliances. And again, once again, very, very low cost. We've upgraded this with the latest Intel servers and that's part of what V5 is about. We've incorporated the latest Intel servers. We've added more flash capacity. We've added software to find networking and faster networking, but maybe the biggest thing we've done is just constantly improved our software and I'll talk about that. So we've been doing these engineered systems for a long time. We are constantly improving and relocating software functions to the right level of the stack to deliver much better reliability, much better performance, and make the thing much easier to use. And a higher degree of automation. The great thing about automation that if it's completely automated, there's no such thing as pilot error. Automation is good for two reasons. One is, saves you a lot of money because labor is expensive. The other is, labor is inconsistent. Software always does the same thing. Hopefully it's not always wrong, but we do a good job of testing the software. You're doing, you know, we always quote, always do the right thing. With human beings, you pay them, you pay them a lot of money and sometimes they make mistakes. Okay. Database appliance. It's lower, you know, again, database appliance includes a compute layer, a networking layer, and a storage layer. Sound familiar? You know, all those, the infrastructure is converged. Compute, network, storage. The network's infinite. And it costs a lot less than what you would buy from the partnership between EMC and Cisco. I want to do it myself. That is really the cheap way to go. Not true. Database appliance costs less than rolling your own using components from Dell. The VCA with FS1 costs more than rolling your own using HP blades and EMC storage. These are, you know, the lowest list prices in the industry combined with the highest degree of automation and the most testing because we engineer and test all the pieces. We integrate all the pieces so you don't have to. Okay, not every appliance is about running faster or even having the lowest purchase price, though this does. The big thing about the zero data loss recovery appliance is that it doesn't lose data. I mean, it sounds funny. Well, of course it doesn't lose data. You telling me, and none of my recovery systems lose data, that's absurd. You know, that's gotta be the case. Really? Okay, so the focus on what's called the zero data loss recovery appliance was to completely automate logging and recovery. Completely automate, take human error out of it, thereby having low cost and never lose any of your data. That's a differentiator because current backup appliances, and there are appliances, I mentioned, that we're not the only ones to have appliances in the marketplace. IBM has Pure Systems, Cisco has UCS, EMC has its own data domain recovery backup appliance. The problem is backup appliances, when you do the restore, when you have to actually, not just, they're great at backing up the data. They take a nice copy of the data and they put it in storage and they save it for you. The problem is, is when you have a failure and you take it off of storage and restore your database. Because if you've done anything between the time you've backed it up and the time you've restored it, whatever you did was lost. Well, that data's all lost. You might have lost hours of data, you might have lost days of data. It depends on when you lasted your backup. These appliances are not very good for backing up databases. They are, again, back one slide please. They lose data. They're actually quite slow because doing the backup while your application is running, it's making a full copy of your database. If you have a big database, that's doing more IOs than your application. It's actually putting more stress on your system to back it up than it is to run your system. And they're very expensive. They are very expensive. You need a lot of them. Okay, so the zero data loss recovery appliance. Again, we engineered it to work with our database. And we engineered it in such a way that, yes, we do periodic backups, but between backups, we take the database logs and transfer them to the recovery appliance. So the recovery appliance doesn't just contain a copy of your database. The recovery appliance was engineered to also maintain copies of the database log, everything you've done since you backed it up. That log entry is sent to the backup appliance. So when you recover, if you have a failure and you have to recover, you can, it can do a restore if necessary, or it can just do a log application or a combination of restore followed by applying the log and you don't lose any data. And in fact, you can recover, let's say it's a malicious attack and someone is inserting false rows into your database. You can do point-in-time recoveries. You can go back an hour and 15 minutes. You can do whatever you want. You have all the information you have. You can go and process your log, take out the malicious transactions and then reapply the log. You can't do that with data domain or any of these backup appliances. This thing was engineered to work with our database. It's fully automated. It has capability to recover, to deal with hacked databases or someone, not really hacked databases. It's just someone in authority. It's usually an inside job. Someone in authority decided to give all their friends a 50% raise the day before they moved to your competitor. So if someone had that authority and maliciously executed that transaction, you can back that transaction out and fix your database. You have a very different, that creates a very different problem if you have data domain rather than the zero data loss recovery appliance. One of these appliances deals with thousands of databases and you can expand the appliance. It scales up linearly. You can back up to your current data center. You can back up to a remote data center. You can back up and log. I keep saying back up. You can back up and log to another data center. You can back up and log to the cloud. So it deals with disaster recovery, the complete loss of the data center, catastrophic loss of the data center. And again, it's faster than data domain, but the big deal is it's fully automated so it's very easy to operate and you never ever lose data. This is about as no brainer an appliance as we have. Okay. We also have a big data appliance. We've upgraded that. Okay, this is our appliance that runs Hadoop and our Oracle's NoSQL database. But we also have a layer of software called BigSQL that allows you to take data out of the Oracle database who we're combined it with. Data from a Hadoop database and a NoSQL database. You can integrate, you can look at all of your data, whether it's stored in a NoSQL database, a SQL relational database. You can take data in an Oracle database and join it up with data in a NoSQL database and join it up with Hadoop data. In other words, you've got a comprehensive three-hundred degree review of all of your data with BigData SQL. And the hardware is the latest, the BigData hardware is the latest Intel processors and all of it is and all the pieces are integrated or are connected with Infiniband networking and the lowest cost storage, okay. And if you compare this, again, to doing it yourself, again, this is the idea of an appliance where all the pieces are engineered to work together. We integrate all the pieces so you don't have to. If you were just going by the parts, going by the parts, it's cheaper than buying all your parts from HP, fans, all your data. Lots of people are buying it and we're not satisfied in that we're adding all sorts of additional software to our BigData appliance, the ability to process BigData. So now we have BigData discovery software, the ability to do analytics on top of Hadoop and NoSQL. That's Oracle. Look at it in a unified way. That's all gravy. Remember, you get all this if you're willing to pay less. But only if you're willing to pay less. Okay, Axilogic, we've had Axilogic in style. It runs our middleware. It runs our applications. It runs any kind of Java application. And again, the interesting thing about this is it's designed as a private cloud. In other words, we run Axilogic, the software that runs in Axilogic is the same software that runs in the Oracle cloud. And I'm gonna come back to this. That's a very important thing. Because when I describe the data center, I'm gonna have a slide at the end. I describe this data center as kind of two components. Kind of the data center core made up of two processor computers. And then a bunch of appliances or engineered systems converged infrastructure. And those two things make up your data center. In fact, there's a third piece of your data center which I'm gonna get to, which is a wire, a wide area network wire out to the public cloud. So we think it's very important actually that what we run on the private side of that wire is the same as what's run in the public cloud side of that wire. And you'll see our engineered systems and our ability to move applications and data back and forth between your data center and the cloud. And what makes that easier is if we run the same software in our engineered systems run in our cloud, like we do with the Axilogic engineered system. Okay, again, it's not surprisingly, it's made up of industry standard components, industry standard two socket servers, infinite band networking, you'll notice some method to this madness. And storage. And of course with X5, it's the latest Intel servers, it's the fastest infinite band available, and so on, and the lowest cost storage. You'll see that with Axilogic, we run much faster, run much faster than the competition, which means our cost performance is better than the competition. All right, and our very first engineered system ever was Exadata, this is the fifth generation of the Exadata machine, the fourth generation that we've built ourself. And it's designed to run all Oracle database workloads, and there are lots of different database load, we're huge data warehouses, there are very fast real-time action processing systems, I mean, very, very different kinds of workloads that the Oracle database has to support. And this is designed to run all of them well. Let's see, it's the latest Intel processors, it's the latest infinite band, the kind of coolest storage, and most importantly, the fifth generation of the software that we've developed for Exadata. All right, it's fault, there's no single point of failure in this system, a compute server can fail, a network connection can fail, a storage server can fail, doesn't matter, keeps running, doesn't matter. The interconnect is infinite band, faster, more reliable. And you start small and you scale out, you start small with a couple of compute servers and three storage servers, and then you just add capacity. It's very interesting that Juan Luisa and I, and Juan's gonna speak after I do, and give you a lot more technical detail, but Luan Luisa and I used to have these huge fights about high-performance disks versus high-capacity disks. And I thought that we should only have high-capacity disks because everything else would be in cash. And Juan says, doesn't matter, people want high-performance. Anyway, Juan won the argument, and we had high-capacity disks. I thought everything would be split between flash and high-capacity disks, which I think are up to four terabytes now. But lo and behold, as with the latest generation, we've gotten rid of our high-performance disk. I'm not sure I can claim this as a victory, though, because we replaced it with flash disk basically for the same price. It's really astonishing. We've thrown out our high-performance disk, we haven't raised prices at all, and give you about the same amount of flash as we used to give you for high-performance disk at the same price. As those prices came together, as the flash prices came together, then finally, high-performance disk became irrelevant. So we have two options now with Exadata Storage, Extreme Flash Storage, and High-Capacity Disk. And High-performance Disk is on its way out, and it's not just in Exadata. I mean, it's fascinating, it took a while. Okay, so, again, flash runs a lot faster than disk. High-performance disks look very high-performance when compared to high-performance disk. Much higher bandwidth, much faster IO rates, dramatically faster system with the fifth generation of Exadata. The software, even more important than the next generation hardware, even more important the fact that we offer you flash and the latest Intel processors are all of the optimizations that we've done. Right now, just how we communicate with InfiniBand, we have a faster protocol tripling three times more efficient than what we had before. That's software optimization. The way we store data in memory has given us, you know, has dramatically improved in our columnar flash cache. Our analytics have been moved to storage servers. Some of the analytics have been moved to storage servers. Again, taking a function that used to be in the compute level and moving it down to the storage level. It's one of the interesting things you can do when you, I mean, that's a particularly fascinating one for me because we really are relocating a function from one level in the engineered system to a different level in the engineered system. And unless you engineer at all, unless you engineer storage and compute, you can't do that. Another presentation sometime in the future, I will talk about moving stuff out of software and into silicon. Same thing. We're doing a lot of it. Trying to put things, always pushing things down, by the way, it's always pushing things down. Closer and closer to the silicon. You can only do it if you work on the compute layer, work on the networking layer, work on the storage layer, and work on the silicon layer. And we happen to do all of those things, which I think gives us a huge advantage, a huge advantage in being able to deliver extreme performance, which we've done for years. And now we're using that engineering to deliver not just extreme performance, but what you could call an extreme list price. Okay. The, we now detect a server death immediately rather than to put in waiting for it to not respond to a certain length of time. The failovers are faster. There are a wide variety of features that we've added to the system to make it much faster, not simply a hardware upgrade, but a hardware upgrade in many levels, storage, networking, and compute, and a software upgrade at many levels as well. Okay, so how fast is it? Really fast, really fast. There's nothing to compare it to. It is, if you look at just the IO capacity, it's much, much faster. You take an analytic load and more in an order, magnitude faster than the fastest thing in the market. We can't find anything that's within a factor of 10, fast, for analytics. Much, and by the way, this is just looking at IO load. The IO load, you know, you're still gonna have something that does the compute. So we are much faster than anything else out there. Much faster. And lower price, by the way, again. Always lower price. We've had a bit of a change in philosophy because of what's happened in the database world. I mean, several things have happened. One is we have a new database technology that allows you to put your entire database in main memory. So, and we did a lot of things to do that. We had better and better compression in columns rather than rows, learn how to process those columns very, very quickly, including using the hardware instructions on the microprocessor. All of those things, a lot of those were software changes. Then on top of that, we now have an in-memory data warehouse looks very, very different than a huge archival data warehouse. And we have to give you configurations for handling these different kind of data warehouses. A big OLTP system looks very, very different than that huge archival data warehouse, and so on. So, we're giving you, for the first time, with Exadata elastic configurations. Where you start with two database servers, back it up please, you start with two database servers and three storage servers, and then you can increment one compute server and one storage server at a time. So, you really can, and configurations. And you can optimize, and I think it's the next slide, you can optimize for the most RAM, most DRAM, if you will, and for in-memory applications, you can optimize for flash, putting all of your data, your entire database on flash, or if your data warehouse is absolutely huge, you combine flash with high capacity disk. So, again, you have the ability now to configure these as you see fit, which I think is a both a blessing and a curse. It means you're gonna have to figure out exactly what your load is, and then optimize your configuration for the load. Okay, and of course, the supercluster, the X5, of course, is the Intel, is our Intel line, we have our processor lines, we have the S line, which is Spark, and we have the X line, which is Intel, and this is our, you know, the Intel generation five, but maybe you know this, maybe you don't, but our superclusters, and we have two different models of supercluster, use the same storage server and the same storage server software as Exadata, and all of the upgrades that we've made to Exadata also go into both of our superclusters. So, including extreme flash, a lot of advanced software optimizations are in places. All right, and this is a slide I promised you, which is it's getting, you know, the next trend, or the next big trend in computing. Everyone's talking about the cloud, but people are not doing a terrific job of connecting. Data centers are not going away, by the way. Clouds are gonna get bigger, clouds are gonna get more popular, and data centers are not gonna go away. And one of the things we're gonna be focusing on is how we interconnect our data centers to the cloud. And I mentioned earlier, you could choose to do your backups, you know, the zero data loss recovery appliance. You could have the zero data loss recovery appliance in the Oracle cloud, and you could do your login backups in the cloud, perfectly reasonable thing to do. You could have some of your access capacity in the cloud and primary capacity in your data center. You could use our cloud in case your machine fails, and you don't wanna have a backup data center. Our cloud could be your backup data center. You could do testing and development in the cloud and do production in your data center. You could do the reverse, you could do test and development in your data center and production in the cloud. But you have to have the ability to press a button and move an application from the cloud to the data center and back again easily and gracefully. You have to have the ability to take a database, take data and move it from your data center into the cloud and back again. So there has to be some degree of compatibility between the public cloud and your data center. And that's another important thing for us, that the things that we run in, the things that we sell you to run in your data center, gracefully connect to things that we deploy in our cloud. So you can, so you can move applications and data back and forth and back again on an as needed basis. And that really is the data center of the future, the combination of all three. The data center core made up of commodity socket, commodity servers surrounded by purpose built engineered systems interconnected to the cloud. That is the future, that is what we're working on. That's what we'll continue to work on. Thank you very much. Ladies and gentlemen, please welcome senior vice president Oracle Systems Technology, Juan Luisa. Okay, welcome. Today I wanna talk about our engineered systems and particularly our database machines. Let's see if we can bring those up on the slides. There we go. So we're introducing our X5 generation of database machines. It is our fifth generation, really sixth generation database machine if you count our original V1 version. And these are all available to order today. And the theme of this introduction is really spending less. So the idea is you spend less by doing more and you spend less by paying less. And I'm gonna go into a fair bit of detail on what these engineered systems are all about, what these exadata is all about. So first of all, a very quick, very high level overview. What is an exadata? It's a platform for running the Oracle database. It has everything you need to run the Oracle database. It has the most advanced scale out computers, networking and storage. It has a lot of special software, which I'll go into a bit of detail on. And it's fully integrated, optimized, debug. It's ready to run. So it's the most integrated system out there. So that's kind of three main components of an exadata. Super fast hardware, super advanced software, fully integrated. Now exadata has been on the market for about seven years now and it's been extremely successful. We have thousands of deployments, about half our OLTP, half our analytics. So we're about even, we're actually getting more OLTP nowadays just because it's a bigger business. We have a lot of very big customers running exadata today. We're running petabyte sized data warehouses. We're running lots of business applications, things like SAP, e-business suite, PeopleSoft, Siebel. We're doing online financial trading. Very, very critical systems. We're running e-commerce sites. We're running lots of consolidated systems. And we're very visible in the public cloud also. So our entire Oracle software as a service runs on exadata and some other very big customers. NetSuite and Salesforce are also very extensive users of exadata. So it's very big presence on the public clouds as well. Now we have a lot of, we've kind of gone through a number of themes in exadata over the years. We started with a big focus on data warehousing, OLTP, then databases as a service. And in the last year, we've been focusing on two new themes which are database and memory and extreme flash. And I'm gonna go into some detail on that also. So first of all, I wanna make clear when we talk about extreme performance, all the time we talk about exadata in terms of extreme performance, how fast it is. Exadata is very famous for being fast. And most people think of that as I can run my workload fast, my query returns quickly, I have sub second response time. And that's all true. But in fact, it's more than that because when we say it's fast, really what we mean, it has more power. So it has more processing power, more computers, more cores, has more storage, more IO's per second, more bandwidth on the network. So what that means is not just your car goes faster but think of it as your car can fit many more people. You can do more work because it's so much faster. So that means you can run more databases. You can consolidate more databases on the same system. You can have many more users. You can run a lot more workloads. You can run a bigger database, a lot more data. You can run a lot more analytics and transactions. So when we talk about exadata being extremely fast, don't think just, okay, I'm gonna get my results a little faster or a lot faster. But think also, you can do a lot more work with the same system. And that's what I call it IT economies of scale. It's extremely important to understand that that as we improve the performance, you can do a lot more with the exact same system. Or you can flip it around and you can say, hey, I just wanna do the exact same thing and I can pay less by getting a much smaller system. So extreme performance helps cost in all dimensions. So I'm gonna talk about hardware first. About exadata hardware, then I'm gonna move on to software. So starting with hardware, the idea of exadata is very simple on the hardware side. We're database optimized, scale out. We have the fastest compute, the fastest networking, the fastest storage, complete high availability. So in terms of details, what's in an exadata? As I mentioned, it's everything you need to run a database. Starting with scale out, two socket servers. We're using the very fastest processors in the industry, the 18 core Intel processors. We're using, we have the very largest memory capacity on those two socket servers. We have the fastest communication network on the internet and in the world, which is infinite bands, it's very modern interconnect. And the thing that distinguishes exadata the most is the storage. Because conventionally you use a storage array for storage and use a separate computer to run your database. In exadata, we had to use computers as storage. So we use two socket servers to run storage. And the reason that we do that is because we're able to put a lot of compute power directly in the storage so we can move database function that processes a lot of data directly in the storage. And I'll get into details on that also. So exadata comes with a huge amount of compute in a single rack we have 168 cores of processors just in the storage processors. Now exadata storage comes in two flavors. There's high capacity storage, which mixes disk and flash for very high performance and high capacity. And there's a brand new version called extreme flash storage which is based on the fastest storage in the industry. And I'll talk about that also. Okay, so let's start with our high capacity storage server. So if you look at a rack of exadata with high capacity storage servers, it's a lot of storage, it's 672 terabytes of storage. And what we've done in our new generation is we've doubled the flash. So we have 672 terabytes of storage with double the flash and we've made it a lot faster. And we've made it a lot faster using three different techniques. One is we're using state-of-the-art NBME flash. So what makes it much faster than other flash? So when we talk about flash and we talk about extreme flash, it's because our flash is not the same as other people's flash, it's much, much faster. It connects directly into the PCI bus which connects directly to CPU. It has the fastest communication, it has no bottlenecks. And we're using the very latest protocols also. It's called NBME, you'll start hearing a lot about that. We're pretty much the first people in the industry to go out with this in volume. NBME, non-volatile memory express. You'll hear a lot about that from other vendors probably a year or two from now when they adopt it. And we've optimized our software also. We've done a lot of work on the InfiniBand protocols to make extreme performance possible. In the high capacity storage server, we tier disk and flash. And what that means is we automatically move data from high capacity disk in the flash so that frequently accessed data comes out of flash to get the performance of flash with the capacity of disk. It's very, very good price performance. And we have all our offload capabilities that we've always had in X-Data. I'll get into more details on that in the software section later on. Now the performance of X-Data has always been amazing and now it's even higher. So now we're talking about analytics scans over 100 gigabytes a second, 140 gigabytes a second. OLTP reads over 4 million IOs per second. These are AK OLTP reads. Writes at 2.69 million extraordinarily high numbers. If you go back a few years, a storage subsystem, a very big storage system might have executed 50,000 at most 100,000 IOs per second. Now out of the box in an X-Data is 4 million IOs per second. Tremendous performance and also very good response time. So we're at a quarter millisecond response time and that's not on an idle system. It's a quarter millisecond response time running a million IOs per second. So extremely high throughput, a very, very busy system still achieves a quarter millisecond response time. It's very unique far beyond what others have in industry. And the only thing that beats that is our extreme flash. So this is our new generation of storage for X-Data. It's called our extreme flash storage server. In this one, we have all NVMe flash. So we have 12.8 terabytes of NVMe flash in each storage server. Again, it's the latest generation of flash. It connects directly into the PCIE bus. It has no disc controllers to slow things down. Because it's all flash, it gives consistent, extremely low response times. It also has all the offload capabilities, everything that we had before. It's very good for IO intensive workloads. It produces very good extremely high performance. But also you can convert that performance into cost saving with large workload. And this replaces our previous high performance disc storage server. A quick drink here. With the latest generation of flash, there's really no longer any need for the high performance disc because we can achieve almost the same level of capacity and have the full performance of flash. And the performance rate there are even higher. So 260 gigabytes per second. So most people that don't kind of live these numbers don't understand these numbers. But that is an extraordinarily high number. Think about if you have 10 gigabit per second interconnects. If you want to move 263 gigabytes per second of data from a storage array to a computer to achieve the similar IO rates, you would have to put 260 or over 210 gigabit per second interconnects between the storage and the computer to do that. And for those of you not in the computer industry, it's not even possible to do that. It's impossible to physically wire up 260, 10 gigabit connections into the storage. It's also impossible to wire them into the computers. And then even if you could do it, your computer would spend all this time just processing all this data that's coming in. So the technique of moving the compute into the storage is the only way to keep up with the performance of flash. Flash has now gotten so fast to this 260 gigabytes per second. It's faster than the network can handle. So the only way to achieve these levels of performance, to keep up with the performance of flash is to move the functionality. It's not possible to do it any other way. And that's what we've done in Exadata. And on this one, the response time, again, quarter millisecond, now at two million iOS per second. So extraordinarily high IO rates, still extremely low response times. All right, and so a comparison. So EMC last year came out with an all flash storage, a brand new redesign designed from scratch to be their fastest all flash storage. It's called Extreme IO. And you can compare a single rack of Exadata against the biggest EMC Extreme IO you can buy, the biggest one that they'll make. So the largest EMC IO against a single rack of Exadata, you'll see the Exadata analytic performance is over an order of magnitude faster, 15 times faster. Even for random IOs, significantly faster, almost three times faster. The response time is more than two times faster in Exadata. The compression is much better because we use our hybrid-column compression. And despite the fact that we're 15 times faster at analytics, three times faster at OLTP, twice or more better response time, it's three times lower cost. So it's much, much better. So this gives you an idea. If you compare the latest and greatest, the most advanced flash technology in the market versus Exadata, Exadata is much better both on the performance and the cost side. And that's not counting all our software optimizations. We have a ton of software optimizations. And Exadata scales. So as you add more racks, we continue to scale. This thing doesn't scale. This is the biggest configuration you can buy. A big change that we're doing in Exadata X5 is we're introducing elastic configurations. So what this means is customers can now expand their Exadata configurations a server at a time. So previously we had standard configurations that started at a quarter rack. You went eighth rack, quarter rack, half rack, full rack. And those were the increments. Now you can expand them at a server at a time. So you start with two database servers and three storage servers. That's kind of minimal, highly available configuration. Once you've done that, you can add either database servers or storage servers one at a time, either the all flash or the high capacity. And then when you fill up your rack, you just put another rack and start adding servers to that. So it expands linearly both on the compute side and the storage side. And that allows the configuration to be sized to the exact needs of a customer. And the ability to do elastic configurations is not confined to our X5 system. For existing Exadata customers, they can also expand elastically their existing configurations. So for example, if a customer needs just more compute power, they can take an X5 storage or compute server and add it to their existing X4 or X3 or X2 Exadata. So we've made this available, not just to the latest and greatest hardware, but also to existing configurations. With the ability to elastically add compute and storage, we can now tailor Exadata for specific workloads. And an important example here is an in-memory workload. So last summer, we introduced our Oracle Database in Memory, which is a new Oracle database technology that runs in-memory workloads with columnar formats, vector processing, extraordinarily fast. And now we can tailor an Exadata to be an Oracle Database in-memory machine. And the way we do that is we add lots of compute servers with maxed out memory. And then we combine that with flash and disk so we can get the best of all worlds. We get extraordinarily fast in-memory processing for frequently accessed data. We get extraordinarily high IO from the PCI flash, both for loading the data into memory and for any data that doesn't fit in memory. And we get high capacity disk for archival data, older data, backups, data that's infrequently used. So you get the best of all worlds with Exadata. You get the cost of disk with the performance of flash and the super high performance of in-memory. It's all combined in one system, an Oracle Database in-memory machine. Now you can also tailor an Exadata for all flash OLTP workloads or for large data warehousing workloads, anything you want. So now it's completely tailorable to an exact workload that you want. And along with updating our Exadata, we're also updating our Supercluster. Supercluster is kind of a close cousin of Exadata. Supercluster is also a database platform. It shares the Exadata storage and the Exadata software and the Exadata technologies. But it also allows consolidation of applications in the exact same configuration. So Supercluster has some capabilities, for example, embedded NAS storage for running virtual machines that allows applications to be consolidated into the exact same hardware that the database is. And Supercluster have a very, very efficient virtual machines, zero overhead virtual machines. They're extremely secure because the virtual machines, the hypervisor equivalent in Supercluster runs in firmware. So it's extraordinarily fast and even more secure than a software hypervisor. In Supercluster, we have extraordinarily high amounts of memory. So in a large scale Supercluster, you can have 32 terabytes of memory per machine. So for running an in-memory workload, that is a huge amount of memory. You can run very, very large databases directly in memory. And also, the memory bus, the memory interconnect in a Supercluster is far, far faster than on a conventional Intel processor. The memory processing in a Supercluster is extraordinarily fast. The virtualization is extraordinarily fast. It's great for consolidating applications in database and for running in-memory workloads. And it includes all this new Exadata technology that I'm talking about. The Xtreme Flash, the new high-performance software features, everything's in there. We also have an 8-socket version of Exadata called the X48. We're also refreshing that and making that elastic. So previously, the X48, you had two 8-socket servers and 14 storage servers. Now, we're making that elastic. We're allowing from two to four compute servers and from three to 14 of the latest generation X5 storage servers. So again, it can be tailored. If you want to use this as an in-memory machine, you can basically shrink the storage and grow the compute and the memory. Okay, so I've talked a lot about hardware. And hardware is very interesting. There's a lot of cool stuff to say about hardware. But the real secret sauce, the thing that really sets Exadata apart isn't the hardware as much. It's the software. Exadata has a lot of software that is very specific, it only runs on Exadata. And this is our biggest software release in Exadata that we've ever had. It has many, many new capabilities and features. The idea of Exadata software is to make the database performance and consolidation much, much faster, make network latency and throughput much, much faster, and make storage throughput latency and capacity much, much faster. Now, first, I'm gonna talk a lot about what we already have. So we've already spent a lot of years building Exadata software. And we have a lot of capabilities. I'm not gonna be able to go into all of them. So I'll just highlight a few of them in here. So Exadata is the fastest OTP platform. Why is that? I talked about the flash storage, the InfiniBand. But what really makes that possible is our special protocols that we use for IO across InfiniBand. We have very customized protocols that enable this. You can't just drop flash hardware into a conventional system and achieve four million IO's per second. It doesn't happen. You'll have bottlenecks all over the place. Only because we tune the entire stack and we have specialized protocols can we achieve four million IO's per second. We have special software for speeding up transactions, for redo log writes, makes them extraordinarily fast in Exadata, makes OTP run much faster. I talked about in our data warehousing, we can run 263 gigabytes per second. That is a gigantic gusher of data coming into the servers. It is comparable to memory performance. So if you look at, for example, standard two socket servers, the memory bandwidth is about 30 to 60 gigabytes per second. So you have to add up a whole bunch of those to get this kind of storage bandwidth. So the Exadata storage bandwidth is so high that it's comparable to memory bandwidth on most systems. It's extraordinarily high and it's only possible because we are able to move database functions directly in the storage so we don't have to move all that data across the network. Exadata also has unbelievable availability. We have a lot of special software for availability. For example, we optimize backups by offloading those in the storage. We have special optimizations for rack interconnect for communication that makes recovery much faster. So we've optimized availability throughout the whole platform, both at the low level, things like disc failures and firmware, but also at the high level, things like backups and cluster communication. Exadata is extremely high capacity system. I've talked about how we have a smart flash cache that sits in front of disc. And the key word there is a smart flash cache. So a lot of storage has a flash cache that sits in front of disc. Ours is smart, ours knows the database. We understand the workload that's running. When an IO comes in to a normal storage subsystem, they don't know what it is. It's just a request for some data. We know what it is. We know this is a backup. This is a report. This is a long running SQL statement. This is a big table it's accessing. This is a small table it's accessing. This is a sort. This is a temp IO. So we distinguish the kinds of IOs. We distinguish the kinds of workloads and we're able to treat the flash differently. We're able to cache intelligently. If you're doing a backup, we're not gonna kick out everything in your flash to put the data in that's just gonna be never used again. We distinguish the different workloads and we are able to do caching much, much faster than on generic storage. In Exadata, we have something called hybrid columnar compression. It's the best compression format in the industry. It typically provides 10 times or more compression for analytics and archival data, which is the bulk of the data out there. Most big data is either analytics or archival. And we have many features that make consolidation better on Exadata. For example, we can prioritize both the CPU, the network, and IO. So when you have a workload, it's not enough to prioritize it at the CPU level if it's gonna get different prioritization at the storage level or at the IO level. In Exadata, because we control the database, the CPUs, the storage, and the network, we can prioritize workload across all the different tiers to achieve the best possible consolidation. Okay, so that's what we've done in the past. Now let's talk about some of the things that are new. What have we done in this release? And I'm gonna give you a quick overview of some of the highlights here. There's more than this even. And then I'm gonna drill down into a few of these. So we have a new InfiniBand protocol that makes OLTP run even faster. We have a new columnar flash cache, which makes analytics run even faster. We're now able to offload JSON and XML. So JSON has become very popular as a data format in web companies. It's become very popular in the industry. So now, as of last summer, we've introduced JSON data types directly into our Oracle database. And in this release, we're able to offload JSON analytics in storage as well as XML analytics. We've done a lot of work on high availability. Instead of waiting for timeouts to detect when a server dies, we can now instantly tell when a server dies because we can ask the InfiniBand interconnect what it sees, what servers it's able to see. And when both switches can no longer see a server, we can immediately declare it dead. We're able to fail over IOs extremely quickly. So for some reason, a storage server is not responding right away. We can fail over the IOs to another storage server in sub-second time. We have very special high availability techniques for a memory database. And I'll get into a little bit more detail on that. We've also added snapshots for test dev environments, for database test dev environments. We allow mixing of flash and high capacity storage, we never did in the past. And we've added virtual machine capabilities, which I'll get into a little more detail on. And we now allow flash to be reserved by different workloads. So when you're consolidating many workloads, we can now reserve a certain amount of flash and the flashcast for each workload so that the busiest workload doesn't take it all away from other workloads that are running in parallel. So I'm gonna drill down into a few of these new software techniques. So first of all, we have a brand new InfiniBand protocol that's even faster than what we had before. So InfiniBand at the hardware level is extremely high throughput and extremely low latency. And we achieved the full InfiniBand throughput when running on exit data. But when sending small messages, we still need to call into the OS for every message and this limits our message rate. So that software overhead, that CPU overhead for small messages limits our message rate. With our new generation of software, because we own the database and we own the firmware and we own the InfiniBand hardware, we're able to do something different. We're able to integrate all that to have the database called directly into the InfiniBand hardware bypassing the OS altogether for all the frequent operations. So we call directly in and that gives us up to three times speed up in OLTP messaging on exit data. This is the first time that this has ever been done. This is the world's first direct to wire InfiniBand protocol and it's in exit data today. Now the thing we've done is we've done a much more intelligent flash caching. So I talked about how our flash caches is intelligent about workloads. We're making it even smarter. So the way flash caching works is you take data that's on disk and you move it into flash so that it can be accessed out of flashes. So flash is faster than disk because flash is faster than disk. Well, we're taking this one step farther. In addition to just moving the data, as we move the data in the flash we transform the data into a different format. So again, this is only possible because we own the database, we own the storage, we own the flash, we own all the algorithms so we can integrate. We can make things happen that nobody else can make happen. So the idea here is if you have an analytic workload, a pure analytic workload, as we move the data from disk to flash we transform that data in flash into a pure columnar format which optimizes for pure analytic workloads. If you have a mix of OLTP and analytic workloads then we can optimize for that instead. So again, this is a big breakthrough in the industry. Nobody's ever done this before. It's the world's first cache that transforms the data as it's loaded into cache. Very different from what other vendors do. Another thing we've added into Exadata is virtual machine technology. So virtual machines are great for providing isolation among different workloads running on the same machine. So isolation in terms of CPU and memory and OS. With Exadata we're able to combine virtualization with our resource management of the network and the IO to provide isolation all the way up and down in the CPU, in the network, and in the storage. So it's very unique to Exadata. Our virtualization in Exadata bypasses the hypervisor, also bypasses the OS when doing IO operations. So we have an extremely fast virtualization implementation in Exadata. It's almost impossible to measure the difference between raw performance and virtualized performance. It's extraordinarily fast. And on Exadata we combine our multi-tenant database with virtualization to get extraordinarily high consolidation density. Far bigger than you can get by just using virtual machines alone. So this is a big new area for Exadata. It's the first virtualized InfiniBand platform. So virtualization has been around for a long time in on the ethernet world. A lot of work by a lot of people in the industry has gone into making ethernet virtualizable. Oracle's done the work on the InfiniBand side. So InfiniBand is a newer, more modern, faster networking technology. And now we can virtualize that also. We can achieve all the performance that we had with InfiniBand in a virtualized consolidated environment. And the last software feature I want to highlight is our in-memory fault tolerance. So last summer we introduced our in-memory database. It's a feature of the Oracle database. Now in-memory databases have one big flaw, which is you need to load the data from storage in the memory. And we're talking a lot of data. So an in-memory database typically has hundreds of gigabytes, even up to terabytes of data in memory. And if a node fails, you can recover that because the data is stored on storage, but you have to reload all that data into memory. So to reload hundreds of gigabytes or a terabyte of data into memory takes a long time. It might take 20 minutes, 40 minutes, an hour to just reload the data into memory. That's a big problem that in-memory databases have. During that time, you're either running really slow or you're not running at all. In most of our competitors, you're not running at all. In our case, you're just running slower. So on Exadata and Supercluster, on our engineered systems, we've done something better. We've introduced something called in-memory fault tolerance, which basically duplicates in-memory data across the nodes. So as a user, you can select specific tables and partitions that you want the data to be duplicated. And after that, it's application transparent. And think of it like mirroring. So if you have a mirror disk subsystem, you have a pair of disks, IOs go to both. If one of the disks fails, automatically all the IOs go to the remaining disk. Well, we're doing the same thing with memory. The memory can be duplicated across the nodes of a cluster. If a node of the cluster dies, we automatically continue to run the in-memory workload off the remaining nodes of the cluster. Because the data is duplicated, there's no slowdown. There's no pause. There's no stopping at all. So the ability to duplicate data in memory is very unique. It's never been done before. Again, this is a new breakthrough that we've introduced now. It's called in-memory fault tolerance. No one else has this. No other in-memory database has it. No other platform has it. Okay, so that's a quick view through some of our new software techniques. Now, I'm gonna enter the last part of it, which is how you spend less by paying less on exadata. Now, there's a few different parts to this. So first of all, you administer less on exadata. There's a lot less administration. Why is that? Because, first of all, deployment time in exadata, it comes pre-built, pre-configured, pre-optimized. Everything's done from database to drivers to firmware to OS to storage and network. Everything's done. You drop it in, you run it. And it's not easy to do that. Typically, customers, even very sophisticated to customers, take weeks to months to deploy a very high-performance database system. You save a huge amount of time on tuning and debugging. When you have all the different components, whether it's all the different storage and flash and this and that, OS, drivers, hypervisor, all this has to be tuned and integrated. We've done that. We've spent a huge amount of time doing that. This is actually a very difficult problem. Anyone that's done it knows this is not easy. We've spent hundreds of engineers tuning, optimizing, and hardening the entire stack. So that's less labor that you need to do. There's also a lot less time spent troubleshooting and finger-pointing. Oracle supports the entire stack, everything. Whether it's a firmware issue, whether it's a driver issue, OS issue, database issue, fail disk, everything you call Oracle. Oracle supports, Oracle takes accountability for any issue that happens in it, anything from database to disk. So there's no more finger-pointing, no more troubleshooting of your own issues. In terms of patching, all these components, when we talk about exadata or anything else, there's a ton of components in there. There's disk drives, there's firmware, disk controls. Everything needs to be patched. Everything needs to be updated. With exadata, we've automated the whole thing. And also, we offer a special support called platinum support. So you don't even do it yourself anymore. Platinum support will do it for you at no extra charge. So when you buy one of our engineered systems, it comes with platinum support. No more patching, platinum support patches the entire system for you, everything. Database, OS, hypervisor, firmware, drivers, everything, storage, networking. Also, exadata changes the management. So typically, when you have a normal system, you have a team that does storage, a team that does networking, a team that does OS, a team that does computers, and a team that does databases. So there's all these different teams that are involved in managing a system. And when you wanna make a change, when you wanna do an update, you have to coordinate all these teams, you have to get everybody lined up, you spend a lot of time talking and going back and forth, and this guy's busy doing that, and that guy's busy doing this. With exadata, most of our customers now have evolved a different management paradigm. They have something called the Database Machine Administrator. So it's one guy or one team that does all the management. And the reason this is possible is because Oracle's done all the work ahead of time. So you don't have to fiddle with a lot of stuff. You don't have to be an expert on firmware and drivers and all that because Oracle's done all that for you. So a single person, a single team can manage the whole thing. All this coordination doesn't need to happen anymore. And the single administrator is aided by having a single management tool. So our Enterprise Manager tool manages everything from top to bottom, everything from the database to the storage to the network. Everything is managed through a single tool. So it's much, much simpler. All these combine to give much less administration. And now I want to talk about one other thing, which is how exadata as a global standard reduces risk. So sometimes people ask me, what's the biggest competitor to exadata? Is it vendor X or vendor Y? Is it this storage or that storage? And actually none of those is really the biggest competitor to exadata. The biggest competitor to exadata is whatever the corporate standard is. So every company, all large companies have developed a corporate standard. It's somebody's compute, somebody's virtualization, somebody's storage, somebody's network, somebody's SAN. They all have their own corporate standards that they've consolidated around. And that's a good thing. Corporate standards are very powerful to help reduce cost and simplify. And they reduce risk. But exadata is different. Exadata doesn't fit into the corporate standard. It's not their compute, it's not their storage, it's not their SAN. It doesn't fit any of the corporate standards. But it's better. And what is the reason for that? Because exadata is essentially a global standard. So important thing to understand about exadata is that every exadata is the same top to bottom. The only thing that's different is the size. So there's small ones, you add more servers, it becomes big. Exadata is exactly the same across all deployments. It's different. That means the full stack is the same. So when we test high availability, we're testing high availability for the full stack, database to disk. When we're testing security, we're testing security for the full stack. When we're testing performance, we're testing performance for the full stack. So when you get an exadata, you're getting a system that's completely tested by Oracle, it's deployed in our cloud, we have exactly the same system in support. And it's exactly the same, not just you and Oracle, but also thousands of other customers. So we have the world's leading telecoms, banks, retailers, et cetera, running Oracle exadata. And so the benefit there is, you're getting not just your own administrators. So what you get with the corporate standard is, you get to share knowledge among your company. When you get exadata, you're sharing knowledge across the entire industry, across the whole globe. So for example, banks really push high availability. So we work very closely with the biggest banks to achieve extremely high availability inside exadata. And once we've worked with them, you get that anytime anybody deploys an exadata, you get that. So you get bank level high availability. Governments are paranoid about security now. There's a lot of hacks going on. You read about them all the time. Governments have become extraordinarily paranoid about security. They look at every detail of every configuration. And we work with them. And so when you get an exadata, you get government level security. Telecoms have very high data volumes and very high transaction volumes. So every call, every click on your phone goes into a database. So telecoms pushes on the scalability side. So when you buy an exadata, you get telecom level scalability. What this means is you're not gonna find common issues. Common issues are gonna be found by somebody else because the system is exactly the same as everybody else. And unique issues are gonna be minimized. So it's very different. Again, this is probably the biggest impediment to our sales. So the biggest competitor to us is exadata versus a corporate standard. And the message here is, what's stronger being the same as everybody in your same company or being the same as thousands of other people in the world and exactly the same as Oracle Development, Oracle Support, Oracle Cloud, it's much more powerful to be the same as everybody else than to be the same as the guy next to you at the same company. And that's really the benefit of exadata. It's a global standard. Okay, so let's talk about spending less by paying less. Very directly, how do you pay less in exadata? First of all, everything I've talked about, the faster processors, the network, the flash storage. All the prices are unchanged or lower than they were in previous generation hardware. So we get much better price performance in this generation. The elastic configurations allow you to save money. So because we can tailor and we can grow server by server, storage by storage, we can design a system that's specifically tailored for a specific workload, whether it's in-memory, database, data warehousing, flash, OTP, can get a much more cost-effective system. And also, as I mentioned, you can take this performance and choose to use it by shrinking your hardware, running less hardware to run the exact same workload. On the software side, there's also a number of savings. In exadata, I mentioned we have virtual machines, but they're not just any virtual machine. There's something called trusted partitions. And what that means is you can license database software by virtual machine. So you no longer need the license, for example, if you have an option to the Oracle database, normally you have to license the entire server with that option. With trusted virtual machines, you can license just the virtual machines that are running that option. Saves licensing costs. The virtualization that we've included in exadata is very low cost. In fact, it costs nothing. So the VM has zero license cost. The VM management tools have zero cost. And the support is included in the existing exadata support. So when you virtualize an exadata, when you go from a physical to a virtual exadata, you do that for no additional cost whatsoever. We have capacity and demand licensing. So if you start with exadata hardware, you can turn off a bunch of the CPUs to basically save on licensing costs. And as your workload grows in the future, whenever you need it, you can enable those cores and only pay for the licensing then. So it's capacity on demand. You grow your capacity when you need it. That's done both within the processors, also by adding servers one at a time. When I talk about running the same workload on a smaller hardware, because it's so much higher performance, that also saves software costs. Because the software costs in exadata are per CPU and per disk. They're not per terabyte. Many other products charge you per terabyte. So as you go to a new generation of storage, for example, you have to pay more software fees because it's larger terabytes. On exadata, it's per disk, not per terabyte. So you don't grow the software costs as you go to a larger configuration. And then another thing that is often missed by customers is that our software licenses are transferable. So in many competitive products, the software and the hardware come in a big bundle. And when your machine gets old, it's four or five years old, then you get a new machine. Basically, you buy a new machine and all the new software. You have to basically pay again for that same software. With exadata and Oracle in general, you can transfer the licenses. So if you have an existing exadata machine, and let's say it's three, four years old, it's time to upgrade, you can buy a new exadata machine and transfer all your software licenses from the old machine to the new machine and pay nothing for the software at all. So again, many competitors, it's a subtle point that's often missed when you're doing evaluations because everyone looks at the initial purchase cost and compares those values. But in fact, three or four years later, with other vendors, you have to basically pay for the software all over again with Oracle and exadata. You can move those existing licenses and it saves a lot on software. So to add a lot up together, a lot of hardware savings, a lot of software savings, a lot of administration savings, lower risk, lower cost from having a global standard. It's a big cost differential. And so that's pretty much it. So that's what I wanted to cover today. The new generation of exadata X5 has a lot of new capabilities, both on the hardware side, extreme flash, faster processors, a lot of new capabilities on the software side, things like our new OTP protocol, virtualization, the columnar flashing, a lot of new enhancements, a lot of new innovations on the software side, a lot of enhancements on the purchasing side also. So very big release for us on all dimensions. And I think you're gonna be very pleased with the results if you adopt this platform. So thank you very much.