Larry Ellison Debuts Oracle Autonomous Transaction Processing

[Music] please welcome Larry Ellison [Music] thank you very much we're we're here this afternoon to talk about the latest developments in our autonomous database technology back at Oracle OpenWorld last year we announced the Oracle autonomous database the world's first and only autonomous database and then in March of this year we delivered the first version of that for production use for data warehousing so this was optimized to run high-performance queries so we could handle data warehouses data Mart's data lakes any any large aggregation of data go in and inquire about that data and return results in an optimal fashion the cool thing about it was because was autonomous the database you know it was fully automated human beings didn't create their database the database is created itself human beings didn't tune the database the database tuned itself we were able to do that for data warehousing and data Mart's back in March of this year and now we're announcing the immediate availability of the Oracle autonomous database for a transaction processing so now the this machine learning based technology not only can optimize itself for queries for database database warehouses and data Mart's but we can also optimize it also optimizes itself for transactions so it can run batch programs reporting Internet of Things simple transactions complex transactions all of it and mixed workloads so between these two systems the system that Optima is optimized for data warehousing and the system that's optimized for transaction processing the Oracle add-on of autonomous database now handles all of your work load all of them okay the auricle autonomous database everything infrastructure the database the data center everything is automated there's nothing to learn and there's nothing to do think about this how hard is it to learn to drive a self-driving car what kind of class do you take - you know who do you hire to teach you to drive a self-driving car how long does it take to learn I mean it's kind of a mean it seems strange but there's nothing to learn if you want to use the oracle autonomous database you know you have to learn nothing so the same thing with a self-driving car now when you get in that self-driving car how hard is it to drive what do you need to do to drive that self-driving well you have to tell youtell where you want to go you tell it where you want to go and then you take a nap or browse browse the web or read a book or whatever so I know it sounds strange nothing to learn nothing to do one of the interesting things about database technology is there are IT experts that specialize in relational databases like Oracle and they learn how to drive these things and the traditional IT developer probably is not an expert at database or the Oracle database or any database in fact one of the reasons no sequel databases are popular is because they're very easy to learn and use it was the simplicity you know you've acted you didn't have to learn a lot in order to use in no sequel database it did a lot less mean it does we have no sequel databases god bless you if you want to use them but the real attraction to the traditional developers using a no sequel database was there wasn't much to learn was get put very simple the Oracle database wasn't you know very complicated to learn how to how to create a database and how to the database into all that that's all gone now now we're as simple to use as the simplest databases on the planet no sequel databases there's nothing to learn there's nothing to do we're much easier to use than any other relational database or object database or a columnar database or what have you we are now the easiest database in the world to use because again I'll say it again nothing to learn nothing to do and when there's nothing to learn and nothing to do it's much lower there's much less labor associated with running this database so it's much much lower in cost in fact it's the lowest cost database to operate ok another reason why the Oracle database is so low cost to operate is that it's truly elastic I know that Amazon calls its cloud the elem the atlast ec2 elastic cloud elastic compute compute cloud but their databases are not elastic let me explain how this works so when you create an Oracle database the system automatically provisions itself in other words it allocates storage it allocates Network capacity it allocates compute capacity in memory or automatically allocates that stuff but then when you start to run your application let's say you're not using all of that capacity you don't really need that capacity it's the load let's say in the middle Knights very low well if the load is very low or a cool start deallocating servers and deallocating servers when Oracle's not running when the when the application isn't running there is no server there's zero servers allocated to the database it's what's called server list cloud or server list system and as you need capacity near the end of a month through the end of a quarter or a busy busy shopping day or what have you it will automatically add servers while the system is still running so as Oracle starts to slow down because more and more people log on and there are more and more transactions it automatically adds another server it automatically adds additional Network IO an additional i/o capacity when it adds the server it will automatically scale itself up as the demands on the system go up so performance is sustained and will automatically scale itself down so when there isn't a lot of demand on the system you're not paying for what you don't use Amazon's databases can't do that they can't dynamically add a server when the system is running they can't dynamically add network capacity they can't they don't can't dynamically take a server away when there's not demand and it's not server 'less when it's idle so this is a case it's a truly elastic system you only pay for the infrastructure that you use that's one reason why we can guarantee to cut your Amazon infrastructure bill in half we're not talking about the total cost of ownership we're not talking about talking about the part yeah yeah we get rid of all the labor no that's not that's in addition this is just the plain infrastructure the cost of compute networking and storage just the infrastructure bill you get from Amazon to run either of their databases Aurora or a redshift just the infrastructure cost we cut that bill in half and of course we eliminate virtually all of the labor which is even even bigger savings truly elastic you only pay for what you use all right last year at open world I did a bunch of benchmarks where I compared the Oracle database running a data warehouse to redshift and there are all of these benchmarks and we were between actually closer to 10 times than five times but we were we consistently between five and ten times faster than Reggie and that was on the identical hardware configurations same hardware configurations we were 5 to 10 times faster well if you're a 5 times faster and you charge by the minute so you know if we do in 6 seconds what they do in a minute we're not just 5 yeah we're just not 10 times faster we're 10 times cheaper because you're charging by the minute these performance advantages translate into dramatically lower costs because you charge by the minute if we run if we if we can do the same amount of work and half the time we're half the price if we can do the same amount and work and a quarter of the time we do we're a quarter of the price and in the case of Amazon data warehouse with redshift vs. Oracle's data warehouse autonomous data warehouse we were between five to ten times faster five to ten times cheaper that's why it's actually very easy for us to guarantee give you a written guarantee that we cut your Amazon bill in half you take your existing redshift application bring it over to Oracle will guarantee that your cost your bill will be cut in half well how did we do in transaction processing that was then this is now how did we do in transaction processing well Amazon's has two separate databases one designed for query processing data warehousing that's redshift they have another one that's designed for online transaction processing and that's called Aurora how did we do against Aurora well in the very simplest case the best we could get out of Aurora you know Aurora at its best is 1/12 is fastest Oracle this is when Aurora is really doing what this is what but I guess Amazon would show up they were showing these slides that Aurora was only here because the second you you have a mixed workload on Aurora we're more you know more than a hundred times faster than Amazon and you can see that our transaction rate with from pure transaction processing went from eighteen thousand three hundred to eighteen thousand one hundred when you added the next workload you saw Amazon degraded quite a bit more you know they have to have one database is pretty good at query processing one great database well I wouldn't say it's not it'll actually doesn't that go to query processing it's ok it's the best thing they've got for query processing and then they got an Aurora is the best thing they got for transaction processing neither one is very good but when you put a mixed workload on one of them when you have a combination of transactions and queries it's really not I mean 100 times slower is not good let me back back up one please can you back up to slide please thank you and so the second line AWS has been around for a decade he had been here for 10 years Amazon Web Services been here for 10 years Amazon runs their business on Oracle now they said they made a big announcement actually they made a big announcement after I said hey Amazon's one of our best references they just gave us another 50 million dollars for the Oracle database and you know the they thought they don't like being our best reference they think of themselves as a competitor so it's kind of embarrassing when Amazon uses Oracle but they want you to use Aurora and redshift and this yeah AWS is great we got Aurora we got redshift what do you use so I don't and don't ask don't ask and then they say oh no but now they make a big announcement we're gonna move by 2020 we think we can get off Oracle by 2020 we think you get off horrible well you know it's interesting they've had 10 years to get off Oracle and they're still on Oracle and it's not going to be easy for them to use their technology it's not going to be cost-effective I mean it's really really hard we've had other competitors no competitors don't like being our best references Amazon doesn't like being our best one of our best references I'm not going to pick on our good customers s ap is another big competitor a decade ago they came up with Hanna their own database and they said they're gonna get off Oracle they're getting everything off of were cool and they're gonna run on Hana ten years later they're still running on Oracle all of their cloud all their cloud applications are running on Oracle virtually all of their customers run on or you know your p customers run on Oracle it's a really good database it's a really good database so good that our competitors our competitors are some of our best references at one point I know there were articles in the newspaper I don't know what Salesforce was thinking but there were articles in the newspaper the Salesforce is going to move off of work on their testing Postgres and they're testing this and they're testing D Beach they're testing all of this stuff no call them they're not getting off of Oracle it is by far the best database in the world and it just got a lot better because now it's autonomous it's not going to be easy first ten years Amazon couldn't get off of Oracle I think we should watch them very closely they got a goal to get off by 2020 s ap couldn't do it Salesforce couldn't do it I don't think they can do it anyway we'll find out okay so next thing not only is Oracle much faster and much lower cost to operate it also protects your data it protects your data against data theft and cyber attacks of any kind of any kind but most importantly against data theft venerable thing about when your data is stolen you sometimes if you're the CEO you get your name on the front in the front page of a newspaper that's not good you don't want your knees on the front page of the newspaper you don't want your data to be stolen the auricle autonomous database automates a great deal of data security we have continuous monitoring to try to detect a threat or some kind of intrusion so we're constantly monitoring this you know to check it for again a threat of vulnerability and intrusion the moment a threat is detected the system has the ability to automatically patch itself while the system is still running there's no system in the world that can do this other than work all automatic threat detection and automatic remediation while the system is still running you know normal systems other systems you have to schedule downtime you gotta take the system down and then patch the system so you have to find all the systems that you have you have to locate all the systems not easy people have thousands or tens of thousands of databases believe it or not large companies but even you know but they're it's very difficult just to keep track of all of them take them down make sure their patch when this is a manual process you know human beings make errors when it's a manual process is a risky process now it's completely automated and it's immediate the system wall running we don't need downtime while the system is running it automatically patches itself and closes a security vulnerability we don't have to take the system down to patch in fact we don't have to take the system down to upgrade from one version of Oracle to the next version of Oracle we don't have to take the system down to scale to add another server to the workload in fact your downtime your downtime should be less than two and a half minutes every month that's what will guarantee again we will guarantee that we have 99.995% availability we'll be including for maintenance and patching you know adding servers everything well then that includes the operating system we can we can patch the patch the OS while the system is running that cludes all downtime everything you're only down two and a half minutes guaranteed Amazon can't do any of this stuff that's why we're at least a hundred times more reliable than their it's a patch to scale the dude to upgrade to do any of that stuff they've got to take the system down we don't we do it while it's running therefore the system is virtually always available a server fails the server fails and your application keeps running this system is fault tolerant the server fails let's say there's a database software bug a database software bug it happens you know and therefore one of the database servers fails but not to be the hardware failed because the software failed no problem your application keeps running we have multiple servers working you know with Oracle Oracle RAC Oracle parallel server we have multiple servers working on it you know supporting your workload so when there's a failure like that the application doesn't even see it transaction just fails therefore automatically application keeps running Amazon can't do that they can't do anything like that everything ever is so we tolerate in other words we don't just because there's a hardware failure the system keeps running we tolerate that failure when there's a software failure a system keeps running we tolerate that failure and because everything is automated the system is autonomous there are no human errors either exit no human accidents and no human malice no mischief no mistakes all goes away because of automation Amazon doesn't have that Amazon's databases - aren't autonomous they can't tolerate hardware failures like that oh so we're certain database server failure they can't tolerate them again there's more reasons four reasons why Oracle is a hundred times more reliable than Amazon we've been working on these problems for a very very long time I'm not going to take you through this slide you know but just to say just even since since version 9 of the database and obviously there are versions that came before that we have been doing all sorts adding all sorts of features and functions to automate the Oracle software to make it on this journey to autonomous we've been working on this for a very long time and finally with 18c it's autonomous and by the way version 19 C which I will talk about right at the end of the presentation is due out end of this year so we have another version coming out very very soon so we continue to work you know we're delivering all sorts of real cool things but with 18 C we are fully autonomous now for data warehousing and transaction processing but it's not sufficient it is not sufficient just to make the database autonomous the database runs on compute and storage and networking someone's got to manage all of that and it can be you it could be us or better yet it could be the system the software the machine learnt machine learning it that this this also could be made autonomous and that's what we've done we've been working on on special you know adding all sorts of capabilities to the software the underlying software and hardware that Oracle runs on operating systems parallel server things like Exadata and you know interconnects all this stuff we do not have made the database autonomous but you have to have the full automated capability on the underlying infrastructure as well and then you're still not done because you have to then automate the processes in the data center so to have a true you know for the have an Oracle autonomous database it's really made up of three separate and big components the database offer itself in the middle but the underlying infrastructure has to be autonomous and the ER and the overall cloud management the overall cloud also has to be autonomous has to be autonomous has to be reliable has to be secure to deliver that those numbers those performance numbers those availability numbers you have to have all three pieces of the puzzle okay how does it work well it automatically provisions itself and we provision it provisions the autonomous database runs on this exadata server and then that's exited is a very interesting architecture I'm not going to go into a lot of detail but it uses Intel Intel processors it uses if you will standard to socket Intel servers but has a very fancy interconnect how they're interconnected it has some very fancy network security built you know in it it has you know some very fancy real life you know reliability soft software and hardware but it we we run on Exadata which is really a parallel server structure and I'll show you that in the next slide so but it runs on our latest fastest autonomous infrastructure and we use rack real application clusters parallel server what it would used to be known as but it's a parallel scale-out system so your workload whenever you're running an application you never have one and only one server running the database you always have at least two servers running the database so if one fails the other one picks up and by the way if one fails and the other one picks up a third a third one is quickly allocated you're never you have no single points of failure in the system the Exadata system is designed so there are no single points of failure still we can tolerate network network losses server you know server losses all of that whether the application keep running it's secure much more secure than other system I mentioned before how it automatically detects and then patches while running vulnerabilities another thing it does you totally unique to Oracle one big problem virtually all databases have is that the tech the techies who create the database the DBAs who create the database can see the user data inside the database maybe you don't want our programmers to know what's inside your database so with Oracle that's not the case with Oracle you know we really have this separation of duties the people who create you know who create the database if you will the DBA is who kind of manage the technology cannot see the user data that it's not true in the Amazon databases the people create their databases can see the user data that's a huge security vulnerability enormous vulnerability that Oracle does not have and of course we encrypt all the data and in that you know both the data at rest and the data and when it's moving in the network again we patch all the software while the system is running not just the database if you can't pat if there what about know there's an OS bug or no less vulnerability you better be able to patch that too so we've been what you know our version of Linux we've been working on people I deal through wise Oracle working on a version of Linux there are eight you know there are lots of versions of Linux why why create one more well we created one more for a variety of reasons performance was one reason security security was another reason we have to be able to patch that software while the system is still running because of all we can patch the database weak otherwise we can't handle handle the vulnerability we've got to be able to do all of it so it's that's why we had to have a automate the database and also automate the underlying infrastructure including the operating system VMs and all that other stuff okay and patch it and obviously manage it automatically all right we we automatically back up the database and it's but you know depending on the policies you set used to the policy you could back it up to the same datacenter you're probably gonna back it up to a different a separate data center we you know keep you know if you if you select active Data Guard we run active Data Guard in another data center and we do that so the system is highly highly reliable and all again all of this stuff is automated the scalability I mentioned earlier you never have you can see this as you need more capacity chunk here comes another server here comes another server here comes another server then you don't need it goes away goes away nobody does this so you only pay for the stuff that you use it truly is elastic there is not nobody but Oracle can do this because of how we do parallel server and how how our architecture works with with rack or parallel server it really is truly pay for what you use it's the system automatically opportunist and optimizes itself so it automatically creates and drops indexes that automatically rebuilds query plans it and you'll you'll see in nineteen see we've added we've added something really cool I don't get ahead of myself but everyone's very excited about this but it we have autonomous regression testing if you will it automatically tests itself to make sure nothing can ever go slower you know as we to new versions of database this is kind of a little inside baseball but but as we make the database faster with new versions and we tune a thousand queries 990 to nine of them one faster one runs slower and someone gets very upset about the one they run slower and that will never happen again now with autonomous with 19c it checks the new query query plans to make sure that if they're not faster we'll use the old query plan it's just autonomous regression there can be no more regressions the system will make sure that ever happens again that's nineteen see you gotta wait till the end of the year for that but but again we continue to opt you know optimize and automate even back to automating testing ok so specialized work lay it loads on automated data warehousing is designed for very complicated queries versus automated tonica's transaction processing it's for OLTP the data is stored differently as the system optimizes itself as we as the machine learns about how to tune itself as machine learning works for the automated data warehouse we use a columnar format the data is stored in column forms with transaction processing it's rows in the case of is some some of the optimization this is just a few a few examples so one of the optimizations for for data warehousing is we'll see if you if you're constantly asking the same query over and over again we'll pre compute the query you know and we'll keep it around so the next time we don't have to compute we can just say you just you just ask that question or someone else just ask that question here's the result you get it instantaneously so you know creates data you know data summaries in the case of OLTP the toughest thing is getting data off of your storage device probably flashed these days not discs but get it from get it from storage and into memory and to do that very fast there's a technique called our DMA so just reading directly into memory not by patent not having a long code path to go through anyway these are examples of performance optimizations for and how they're different between transaction processing and data warehousing and now the machine learning the Machine has learned how to do both with the latest version of the Oracle database okay there are lots of benefits you don't we upgrade the OS for you at the OS has of the security vulnerability we patch it for you automatically we keep it up-to-date so you know that your version of the database and your version of the OS and your version of the VM have all of the latest security patches on them and that we're constantly monitoring for new vulnerabilities and if they one shows up we immediately patch that there's no delay and we do that across the board so I know some people say oh my god all this is automated what what yeah I'm a DBA I've been an Oracle expert for years am i you know in my upper early retirement I got some bad news for you know early retirement your job I mean a lot of the tedious things you were doing allocating dis storage and you know upgrading the OS is you don't have to do that however there's lots more to do we you be able to work with the developers to you know create new applications improve the analytics may I do you do a better job with data design I mean CIOs are called chief information officers you know because they're trying to get the most out of the company's information DBA is should be helping you know our database administrators to administer the data they should be marching exactly in the same direction trying to maximize the value of the data create more applications improve the analytics improve the insights available the company on the data they shouldn't just be patching the operating system they shouldn't just be adding more storage buying more storage and doing all of that it's nice you to automate the tedious stuff away so you can focus on the mission which is to get the most out of your data and build new applications for your customers and of course by not having human beings doing some of the tedious clerical work in terms of running the database if you eliminate human labor you also eliminate human error and your system is much safer much more reliable much more secure okay I mentioned this earlier developers who are not database experts now can use the Oracle database it's now the easiest database in the world to use so a developer does is not dependent on a database expert to set is set the database up for them and they you know they can do it they can get started right away they can create their own database they don't have to rely on other people to tune to tune their application the application will tune itself and this makes developers and IT organizations dramatically more productive and that's a huge benefit I mean productivity being able to your applications out sooner those applications being more reliable more performant having better you eyes all of that stuff is made possible by using the most advanced autonomous database and it's very easy by the way if you're if you're using a more Amazon Arora and we encourage you to do this to go from Amazon Arora to move over to Oracle autonomous database or as your sequel microwave doesn't mean you know Microsoft's product or it's very easy to simply use our migration tools to pump the data from these manual databases into the Oracle autonomous database it's fascinating most people who are breached most people who suffer data theft suffer data theft long after the vulnerability was known about and a patch was available to fix the vulnerability I mean it's and that makes it all the more embarrassing and costly so the great thing about the Oracle autonomous database the second of all nura bility is detected the system patches finds all the versions of the database they're vulnerable and immediately patches them there's no time delay and you know they've all been patched because the system patched them you weren't relying you weren't reliant upon some manual process to schedule the appropriate down time to get your database secured the detection of the vulnerability is automatic the remediation the patching is automatic and immediate okay you'll see the number 99.995 all over the place it's highly highly available failures we tolerate hardware failures we tolerate software failures we eliminate human errors the database you get a new version of Oracle so let's say you know where I said we're on 18 C now or the autonomous database out at the end come as out then the year is 19 see and okay well how much downtime do I have to schedule to move from 18 seeded 19 see I've got to change all the database offer how much downtime none none because we have multiple servers running your application multiple database servers what happens is let's say you let's say you only have two you could have 2 4 16 whatever but let's say you have the minimum you have to running on exadata what happens is we take one down upgrade this offer from 18 to 19 bring it back up take the other one down upgrade from 18 and 19 bring it back up and you're running so what happens is for a little while you're running on one server actually that's not even that's not true because you have another one in reserve you have you you you're always fault tolerant so you so if during that upgrade there's a failure and the server you're still running you'll fail over to another 18 C server so you can argue there's always there's always at least you know conceptually there's always at least three servers when you when we're doing the upgrade but we can upgrade change this offer well the application still running no schedule downtime the system constantly is tuning itself this is another thing so so it's not typically when an application is built by a developer and then the database expert goes and it Tunes the application once it's tuned it goes live it's tuned in its life after a while a week months a couple years the database might have gotten a lot larger the distribution of the data could have changed the tuning that was done initially might not be appropriate for the current situation there could be more users more data different distribution of data all this other stuff the Oracle autonomous database continuously Tunes itself it's not like one-off in the beginning then maybe once every year or something like that it's constant continuously tuning itself this makes it faster and much cheaper to operate because we're yeah because we're optimizing both for performance and using a minimum amount of infrastructure to get that performance and you just pay for the infrastructure you use we even handle we even protect against user errors let's say someone by mistake deletes you know a thousand customers all the beginning well you know last names begin with E I think like Amazon will probably delete my account I better today I'll probably fortunately I have a guy I just bought the latest Kindle so it's great product so if someone makes a mistake and deletes data they shouldn't have deleted we have this this thing called flashback queries that allows you to go back go back to earlier versions of the database not backups you know earlier you know earlier versions of a specific query result and recover that data they don't happen to anything like that so we have the ability to even protect against user errors whether they're mr. honest mistakes or done with malicious intent we handle all of those situations makes us much more reliable much more secure and much more available okay again I think I said this before very easy to move an Oracle database into the cloud an on-premise database into the cloud or move my sequel or Postgres or you know something else into the cloud you simply use our data pump which will make the appropriate data format conversions make sure the data is encrypted remove administrative privileges that create security vulnerabilities does all of the all of those things and the fact and we use golden gates our golden gate technology to make sure even during the migration your system is accessible and your system can keep running okay so the lowest cost database to run if you're an existing Oracle customer if you already have Oracle licenses you don't have to buy any Oracle database offer you already own Oracle database software so if you have on-premise licenses you can reuse those like there's nothing to buy there's no salt no database offer to buy you just reuse those licenses in our cloud you continue to pay support but those database licenses are now in the cloud they're not on-premise so there's no additional software costs all you pay for is the infrastructure that you use and again we were very good at minimizing the amount of administra infrastructure you used because we automatically scale up and scale down it truly is elastic and because of the that elasticity you only pay pay for what you use we can reduce your costs dramatically I know we guarantee 50 percent we guarantee 50 percent we can do a lot better than that our very incision we we cut costs a lot more than 50% just on infrastructure alone and then the the the fact that we eliminate the labor that we eliminate the labor is even a bigger deal I'll go to the next slide I mean this is really more important than cutting your Amazon infrastructure bill by 50% or 60% or 70% or 80% whatever it is it's gonna be more than 50 for sure but the biggest deal is all of that labor this now goes to zero all of that cost of and all of those valuable human resources all of those experts as database experts can now be working on building new applications building creating new databases getting better improving your analytics getting better insights about your business those precious resources which are in short supply can be redeployed from a more useful tasks than allocating storage and allocating but in and buying network cards and doing a bind and those kind of things there's huge amount of cost in down time when you when you have to it can be extraordinary cost if you have to schedule downtime to patch a security vulnerability and you your data is still in between the time you schedule the downtime and you and you did the patch the costs can be extraordinary but just downtime in general downtime for maintenance downtime you know downtown because of hardware failures downtown's good with software failures down times because of patching all of that downtime is very expensive it has to be scheduled there's a lot of Coast's costs associated with it those costs go away there are no costs and of course the productivity back to the productivity of your people who are no longer doing administrative tasks but again are building the next generation application with the next generation voice UI with the next generation visualization analytics freeing up the resources to innovate rather than administrate makes us makes all of all of the users more more effective okay you can try it for free it's an it's immediately available just log on log on to our cloud and get a and and give it a test drive I think you'll like it the okay I take I'm gonna almost done I'm going to take two minutes I mean everything I've talked about so far is available immediately we have it autonomous transaction processing autonomous data warehouse all of this stuff is available now I'm gonna do a little bit of a teaser I'll talk about this at Oracle OpenWorld and that's a new version Oracle 19 C which I think will be out before it's supposed to be out in 2019 but we've already named at 19 C but it might be out in 2018 I hope you forgive us so you know it's gonna be out either at the end of this year or worst-case January next year one of the coolest things you don't have to read the slide one of the coolest things is this idea that as you upgrade to 19 C there will be no performance regressions no query plan performance regressions so we think all your queries will run faster except the ones that run the same and that's the worst case is running the set because well literally as we run the application run those queries for the first time will kind of run side but you know we'll run kind of a test side by side of those things the old query plan the new query plan of the new query plans faster great you get the new query plan the old query plan is faster you keep the old query plan no more performance regression testing that's a huge effort that our our customers have to make to upgrade from one between one version of our database than another is all this all this testing for regressions for you know where the system has gotten worse not better system does that at four you know in 19 see how good how good is our automated tuning I mean interesting question with 19 see we did you know we did they're interesting tests we took an enormous ly complicated system with high quality experts tuning it over a long period of time and that's the NetSuite ERP system the NetSuite cloud ERP system and actually straight out of the box right now current state of 19c straight of the box is slightly faster slightly faster than what all of the experts were able to do over the last 20 years so this has had 20 years of expert tuning this is their business making this system run fast and they've had lots of experience with lots of different use I mean they have thousands and thousands of companies on this thing I think 20,000 companies to me like I'm using the CR P system and already the expert system is a little better than the 20 years of expert tuning that's gone into it it's kind of remarkable and it's and it just and the machines just gonna get smarter and smarter and smarter the other thing we're offering with 19c is the ability give you give you a choice between what I've described in 18 C as the server list system again the server list system is you go ahead and you provision the system but when the systems are not running you have you haven't reserved anything specifically for you you're on this in this pool of Exadata machines again it's a cloud there are other tenants in the cloud it's a multi-tenant system there are their tenants in the cloud so when you're your application is running you're using some member of those Exadata servers and when your application is not running you're using 0 none of those servers and you only pay for what you use and we think a lot of people want exactly that it automatically scales up and scales down I think a lot of people want that however we think some of our largest customers are gonna say yeah you know I don't want us cheer I don't want to share a my Exadata with anybody I don't want any other companies on this machine or on this network I don't want any noisy neighbors I don't want anyone near me I want to rent the entire neighborhood myself when I'm not in my house don't read it don't put it on the air B&B I own it keep everyone out no one touch it so we give you we give the ability to have dedicated infrastructure say ok I want to get five exited machines or 20 exited machines then I want them absolutely dedicated to my bank or my phone company or what or what have you and then our customers then have the ability to partition up their their workloads using our container database and there are all sorts of things they can do but then but then it belongs only to that phone company only to that to that bank and you're not even sharing the network you don't even share that you know the network between the the computer the the compute system and the storage system you're the only one on that part of the network it is dedicated to you the network is the storage is the compute is you can partition to the other the way you want nobody else no other cloud offers this kind of isolation inside of their public cloud so again we give you giving these two choices you can have the multi tenant system where you're really optimizing for lowest cost in other words if you're not using the server let someone else use it okay you know I'm fine with me and I think most of our customers will kind of end up that way but I think some of our biggest customers will like this idea I've absolutely dedicated isolated that's my storage that's my network doesn't that's my compute I'm the only one using that but you got a choice we offer both and okay and that's what I just said and I think this is pretty much my last slide and cloud and customer so some of our again some of our very very large customers because of statutory requirements really don't want to be in a public cloud at all even the even the dedicated infrastructure which is pretty isolated they just they want all this stuff behind their own fire and we okay we can basically build the Oracle cloud in your data center we can put the Oracle cloud in your data center and you get the same autonomous capabilities the same you know if you got the same OS the same VM you get the same hardware II do you know the exit data hardware you get this everything is the same same software so we can deliver the Oracle autonomous database all the automatic patching all the reliability security availability and we can deliver that inside of your own data center it's the same business model you don't buy the hardware you know you you subscribe you subscribe to the hardware you subscribe to the service it you just say no no I don't want it in your cloud I want it I want it in my data center I mean there are there significant advantage that you you have a very high speed network so there's there's very low it's very low latency high speed network because it's right in your data center and again some of some very big customers you know again big phone companies big banks are highly regulated and they want that additional level of protection they want all of the automation in a cloud they want the otoño autonomous capabilities of our database they want the security they want the reliability but they say can't you just stick that on our floor rather than in your public cloud and we so we take the exactly we basically take a little Oracle cloud and build it into their data center and that's unique to Oracle we're the only ones that are doing that but our very largest customers have said that's the only way they're able to go to the cloud now so we're making making that available and that comes with 19c so in conclusion the Oracle autonomous database takes this manual process of creating and managing databases and automates the entire thing end-to-end this gives you a much more reliable system a much more secure system a system that protects it protects against data theft the system that's up nineteen ninety nine point nine nine five percent of the time and a system that makes you and your developers dramatically more productive thank you very much [Applause] you

Channel: Oracle

Views: 38,628

Keywords: Artificial Intelligence, oracle autonomous database, autonomous database, Larry Ellison, Database, 30669255, OLTP, machine learning, EXT-18-211, Autonomous Database, search-gallery, Autonomous, retire-2018, ellison, English, self-driving, database, self-securing, self-repairing, Oracle Autonomous Transaction Processing, Autonomous Transaction Processing, Keynotes and Presentations, AI, LJE, transaction processing

Id: cJxmowxEsRA

Length: 54min 28sec (3268 seconds)

Published: Tue Aug 07 2018