WWV2016 Amazon Web Services David de Santiago Digital Innovation and Disruption: The Amazon recipe

today what I want you to out it’s more of the actual concept around disruption and innovation so I’m sure that you know is one of the hottest topic right now I’m sure everything everybody every one of you and your company are trying to find out okay what does this disrupted doing how can I innovate at the same pace or how can I be more innovative than the others and you know at the other day you know when you think about innovation what really changed over the past of the five years is that the same concept of innovation is actually becoming co2 the strategy of the companies it’s not anymore a nice to have and to highlight a point what I will do is that I will talk about the Amazon way of doing it the Amazon recipe was a secret sauce that Amazon is doing is using to actually innovate and disrupt so quickly so just a word of a word of cautious here so I’m actually I actually work for amazon web services right that’s not the same thing that amazon.com the retailer Amazon Web Services is the public cloud amazon.com the retailer that’s where we spend our money that’s two different things so just a state of the nation here state of the retail nation before I start with this concept of innovation what’s really staggering when I talk to my customers is actually you know the fact that this race to us innovation this the pace is actually increasing so it’s not only a technological files I mean if you think about you know just five years ago the predictions were saying that one dollar in one in twenty dollar will be spent online by 2020 the reality is that next year one dollar in 10 will be spent online right another concept here that I like to to to to talk about is the fact that it’s irreversible you know the irreversibility of this of this officer of this trend you know for the first time last year when you think about black friday when you think about the 11th of December in China for the first time last year we experienced a decrease in sales two weeks before these events so effectively a digitally driven event has made me change the way that I actually shop and the other concept is the survival aspect here so I’m not going to talk about you know the fact that disrupter are disrupting the place obviously you know another thing to another thing that you know we take lightly here is the fact that you know again for the third consecutive year in the US and now is happening here in amia in Europe the number of physical shops are decreasing you know think about it that’s an economic i would say ethos that we had for more than a century if you build they will come that does not work anymore and when i talk about the ship from physical to digital the reality is now is changing is not so much of a shift anymore is more of a convergence right what you see is that the digital world is being embedded now into the physical world and this is the resultant of all the omni-channel customer centric strategies effectively is the shop that is reinventing himself from a technical technological standpoint of course there is disruption when it comes to technology if you think about you know just mobile the mobile aspect of it if you think about the line of business applications by 2017 every single line of field based application will be mobile effectively what it means it means this is a translation of the API the sharing economy of course everything is going mobile but what’s the problem with that the problem with that is an a mobile application as an average we make six seven API calls half of these LPF calls are IPS that you do not own huge problem for integrating that data that data is actually growing if you think about the data that we generate over the course of the last few years right it’s ninety percent of the level of the data has been generated over the past two years next year they already saying that will be generating more data than last year the prime with data that it creates gravity so when you think again the Amazon the gaffer the tunas all the disruptors the thing that they have in mind and that’s what makes them extremely different from the others is that when they look at data they think that data is capital they treat data as an economic asset of production they don’t throw it away they actually keep it and then mine it and they call that mining the mind the more in mind data the more data you generate that actually refines your algorithms and that’s how they operate right if you think of all the disruptors of the world

that’s exactly what they are doing they are treating data as capital the third thing I want you to talk about as you need to mind the data you need compute power you need somehow you know computers that are actually crunching the data so you see a great move of the compute analytic power of this world to as the cloud and because you need to be more clever more prescriptive more predictive right behind recommendations behind the scenes that you have it’s tons and tons and tons of algorithms and machine learning mechanisms all these mechanisms are no this crunching is going into the cloud eighty percent of the workloads will be data-intensive it’s not anymore the case of saying up what I put in the cloud is my web-site why because I need elasticity right now the core of your business is actually going into the cloud and I’m not even touching on IOT but I OT especially in the world of retail with it will be certainly the biggest disrupter so it’s not the fact that we have sensors right that are able to tell you a lot of things about movement tracking and all that kind of stuff the main thing about IOT is basically that it makes the concept of Internet of now a reality the Internet of now is all about understanding the context of the intent it gives you a location a movement you can derive intent and that’s the main thing about Internet of Things and you can imagine the pressure that will will put on IT organizations when it comes to actually understand that so from a business strategy perspective what you see in companies right now is that there is this ethos that you know you need to move for the traditional sustainable competitive advantage model which is a model that we had for the past 50 years amid all that consists of saying I’ve got different factors variables in my in my business if I somehow can control of these variables then great you know I will find myself in a competitive advantage situation but think about one thing here is that you know what happened the fast in the past 10 years is that consciously we put the customers on the driving seat so nowadays is the customers that decides to how the what the when and the where so you are not in control whatsoever what customers want from you this model does not work anymore you can take the example of the disruptors disruptors they move from that model to a model which consists of continuously innovating the model of continuous innovation that’s what they do at cool that’s why they exist when you think about Amazon we were discussing this at the beginning amazon started and studied by selling books right online but if you think about it you know amazoncom the retailer primarily picked the book industry because it was an industry where the value chain was so big that it was actually easy to cut down the chain so it was easy to actually reduce the number of steps in between there for what they are doing is that going for an industry that was easily transformable but it’s not because the ethos was selling book the ethos was to become a digital company that could actually reproduce a model in other in other areas amazon.com in essence is a technological company technological companies that go and disrupt industry that are based on all models that can be digitized I can provide where you can provide better customer experiences that’s how Amazon works that’s what ueber works that’s how all the a b and b’s and all the guys that you you know you hear on the news now actually t-strap into this work they reduce the value chain and they create better experiences for your customers so just you know how did amazon get into you know cloud computing so how come a company the starting selling books ended up being the leading force of the public cloud and the answer is release it that’s what I mentioned right they were looking at industry that ready to be transformed industry that we are working on all models so as the company expanded they created an infrastructure to support the theory so they created data centers all over the world that will enable wear them to be agile to be quick to be reactive to be more customer-centric and in 2006 they realized that actually you know they became fairly good at building this type of infrastructure so they decided actually wide not commercializing this infrastructure the clever thing that they did is that actually created a completely separated company AWS Amazon Web Services and what’s important to understand is nowadays you know is two different entities but the key thing is that actually for me amazon is one of my customers actually I talked to the guys at amazon.com and they are one of my customers we ran on two different infrastructures they’ve got their infrastructure we’ve got our own infrastructure and the main mission here

is to enable you know businesses and developers to create sophisticated modern scalable and secure applications using web services using web services web services is basically the techie term that we had few years ago to talk about the cloud amazon.com is my customer you know that’s something that you know I want to beat you to be clear here and if you’re familiar with the IT world you know there is a thing called The Magic Quadrant from Ghana it’s kind of like the Bible for the techies basically you know kind of benchmarks and ranks all the different vendors that are doing that are selling something within a given category so this one is the gather for infrastructure as a service the public cloud one of the interesting thing here so you don’t effectively what were you want to be is on top right corner right the vision and the ability to execute that’s where Amazon is but the most staggering thing here but about this this quadrant is effectively what they are saying is that if you combine all the other players together and you multiply them by 10 you get the size of Amazon of AWS Amazon Web Services all of the others combined by 10 combined x 10 to put things in perspective here so imagine either Amazon where it was a seven billion dollar company right that was few ago effectively every day we are adding enough server capacity to power a company of that size every single day it’s staggering the growth in that market is absolutely staggering I’m not going to get into the detail of this one but if you think about the public cloud just back in 2006 2007 2008 effectively the crowd was storage computers and you if you were a bit adventures you might have databases wow that’s what you had few years ago nowadays with the pace of innovation that we have we’ve got companies like Conde Nast like many others that are actually saying I will close my entire data centers I will not have data centers my IT department will run completely on the SS from the network to the deployment of the tools to analytics to compute power everything CDN CloudFront whatever you want all these services are now available on AWS effectively you can have a virtual data center running on AWS and that’s where the market is going and the last slide about you know AWS because I want to get back into the innovation topic one of the key things here when you when you look at a cloud in general the public cloud in particular is the global footprint right you know I’ve got customers in the u.s. in the UK here now in the Netherlands not actually looking at opening operations somewhere else I want you open operations in Australia I want your preparations in Asia using Singapore effectively you know when you operate in the cloud you’re talking to an API so give you an example you know the shoe Scott Geiger they open operations in Asia in three now in three hours effectively when you’ve got an in and an infrastructure that is running in Europe let’s say that you are in Dublin or you’re in Frankfurt changing a parameter that says copy paste or you know delete Europe put Asia minutes later you’ve got the exact same infrastructure using running on running in Asia many many many retailers are looking at the global food at the main thing to get into into the cloud kurt geiger many many others so we have an announcement fairly soon about one of the major retailers here in the netherlands as well so getting back into you know there’s a concept you know of innovation and we know what’s the recipe around that so the main thing to say about innovation effectively is hard right you know it’s not easy to know in advance what you customer want and that’s a main problem about innovation what what do i have to innovate one of the code that i liked you to to to to mention here is the one from a from henry ford when he created the Model T you know if I would have had my customers in advance what they wanted they would that told me fastest faster horses and that’s one of the key things that we hear about our customers is that you know they don’t really think about you know how to transform themselves it makes it even more difficult especially in the retail world when you’ve got so many touch points where do I start to a start in the physical world i stuck in the digital world who has structure to strike a balance in the middle that makes it extremely difficult i was argued actually in the retail world the concept on innovation is harder because of the of all the touch points that you have so you’ve got two options in this situation right i need to innovate right is not it’s a survival for me right you know effectively when you think about it clothes shops are actually closing in some cases so you know first option so you know the ostrich type of scenario

bury my head in the sand I don’t don’t do anything well the reality is that you know this will have consequences just from a web perspective if you actually do nothing what happens is that if you provide poor experiences on your website right the page rose is really slow effectively you will lose a transaction I would say that you know at the end of day then at the end of the world you lose a transaction the probably in the transaction and I think of me and I’m sure you could recognize yourself series if the website sucks right he actually screw me once it not going to screw me twice I’m not going back into that website and that’s what this survey is actually showing and the third thing is that you know when I get when I have a good experience on the website will I be boasting about it but if I’ve got too bad experience on the website then I will be actually you know slashing that website online I’m more likely to talk about it and the most important thing here is the brand damage that you get that’s from a web front perspective let’s talk about the data right being more prescriptive understanding better your customers the customer centric view effectively when you think about it here is more of an icy problem why are we not prescriptive why are we not more why are we not smart about with your data is effectively the lack of technology the licensing problem that we have the fact that IT departments are really rigid they’re not age I and this has two consequences is that effectively you know you talk to the same old data and you talk on the some of the data while the data available to you is growing if you think about traditional enterprises in the retail space traditionally they only use twenty percent of the data that they have to make the decisions within the data warehouse system only twenty percent of the data and data ends up here and that’s what you use to make decisions that’s what is changing that’s what the disruptors are completely changing they are getting data that they have and they are trying to get data that they do not have as well put things in perspective in real numbers so when you think about amazon.com right globally worldwide hundred milliseconds delay cost us one percent of sales at the global scale of Amazon one percent of sales equates more or less to my monthly salary I’m joking absolutely not it’s a lot a lot of money right so you know the consequences are real and you know I talked about the concept of survival I like to quote Vince Lombardi here and I see that increasingly you know if you are not fired with enthusiasm believe me you’re going to be fired with enthusiasm we are seeing that increasing lands increasingly in not innovating is not anymore an option and when you think about effectively you know what amazon is doing well how man wasn’t thinking about you know innovation and the digital and the digital backbone to unleash this innovation is effective effectively this ethos of saying I need to move away from where people behave the way computers work which is the old where she was a world where computers work the way people behave that’s what customer-centric means and having a digital platform that actually enables to do that so how do they do that three simple steps and I’m oversimplifying okay I’m really oversimplifying but just to give you an idea of how we are organized on how we work the first thing that we do is that we experiment frequently in every single business so that could be a WS that could be amazon.com that could be kindled that could be prime that could be media the media agency all of them the experiment frequently I would argue that they don’t experiment frequently they experiment always it’s a live laboratory the second thing that they do is great to test right obviously you know innovation is hard what do I need to put in front of my customers let’s test a lot of things the key thing here is actually to be very practical pragmatic in terms of how you testings measure measure measure measure everything in fact at amazon all the businesses together when we release a feature and let’s say that you know it’s got 10,000 lines of codes half of the line of course actually lines that are actually guilty was measuring the impact that measuring the adoption pattern of that feature so we don’t put so much effort in actually the features we are putting the effort in listening and understanding how this feature has been adopted and the third thing which is kind of a resultant of the toughest things is focus on your business and focusing on your business i will show you a couple of examples is effectively you know the content of saying as a retailer am I an IT organization do I need to run datacenters do I need to have people actually you know just

handling database backups is it different changing me from the others and I will show you how people like actually getting away from that well and AWS wrapping up all this is the digital platform digital platform exactly the same concept at Airbnb exactly the circumcenter exactly the same concept at all the retailers online that are being successful right now they’re using digital platforms thats why they exist experiment frequently the prime with experimentation is basically cost of failure if you don’t experiment is not because you are stupid you are all a really clever bunch of people believe me the problem is actually cost of failure how much it cost you to actually test something it’s not because you don’t have the ideas the most staggering thing I get when I’ve actually visit my customers is this single is this single file that telling me now that I am using the cloud effectively I can launch 10 ideas I test 10 things and I’ve got the strength of two weeks to test it I know that nine ideas we actually fail I’m fine with that because the idea that will work might be worse might be worth millions of dollars in the old world you cannot do that you need to make sure that you just test one idea and you pray that this idea actually works how you do that to be honest with you here is just a cool benefits of what the cloud gives you an elastic capacity let’s take the example of black friday how many customers i see that are actually telling me that on black friday the website was done because of traffic that should not happen that should not happen elastic capacity is the context of saying if two day I’ve got a traffic that requires 10 servers and for black friday I’ve got the traffic that requires a 10,000 servers is completely transparent to me that’s what you get on the crowd auto-scaling you actually follow the curve all your consumption and you just provision what you need right that’s what you get on the cloud at the really beginning five years ago to be honest with you retailers adopted the cloud for that simple reason because they needed elastic capacity because the choices is simple if I need for Christmas 50,000 servers I am the investor I am actually one of the members of the board members of the company I’ve got the choice of saying fifty thousand seven for Christmas I build the data center with 50,000 servers what do I do with these 50,000 servers on the first of January because you know i’m just using 300 of them and effectively what happen is that you strike a balance in the middle and you say i’m going to get 20,000 servers and pray for Christmas as effectively what you sing on the market this is completely changing with the dynamics of the cloud the second thing is obviously you know if you need to invest on servers if you have to make heavy upfront investment is not going to work right you know we are all shot in cash so you know the concept that is no capex everything is operational opec’s the concept that is only pay as you go you pay by the hour right so you know I’ve got customers that are running monte carlo simulations are running different simulations to actually understand the customer patterns in the shop effectively they might run a cluster of ten thousand servers they ran it for two hours and whether they’re done with it they are back to zero servers they pay for 50,000 servers or how many servers just for two hours the key concept you understand here is that if you run 10,000 servers for one hour it costs you exactly the same thing than one hour of 10,000 servers you are paying exactly the same thing 10,000 servers one hour 10 20 10,000 hours one server obviously in the cloud you can do things much quicker if you can throw capacity to it and we actually peg the end up paying the same and available on demand so this is the concept that you know when you need capacity you know ideally you don’t want you have to wait for the truck to come and you don’t have to wait six months to get your your capacity or do capacity augmented is effectively I 150 thousand servers I want them now it’s a click you’ve got them you can use them a minute later that’s how you lower risk that’s how you lower the cost of failure measure improve iterate so this one is my favorite one measure improve its rate so that’s effectively what amazon.com in my opinion does extremely well and I will put that in the context of a few example like a quiz imagine a graph with two axes and axes with this scale and the other axis which is the time scale is effectively saying okay the adoption set rates are here customer like my application that’s when the things takes off right that’s the end of the curve if you and time if you want is move the console saying I’m testing things you know I’ve got an idea I will test this and measure Sid see if this my

idea takes off and if it doesn’t take off I analyze what what what my customers are doing and rewrite the feature to actually behave in a different way that is more suited of what I’ve seen previously that’s what all these Raptors are doing when they launched products let me give an example here so a company that started with the name of bourbon and their initial idea was a location-based mobile app and photo sharing is what just one of the features that’s effectively we are here bourbon starts I’m creating something that is just a location-based mobile application this company now is Instagram they went through six iterations understanding the customer feedback understanding the how the product has been adopted rewriting rewriting rewriting in very short periods of time very cost-effective and when it takes off it takes off he takes off big another one here and I started with the name of o’dale and it was basically a put a podcast sharing application that was the initial idea the first spike that’s Twitter that’s exactly how they started I can tell you that all the disrupter are starting right now every single feature that you put on the website in amazon.com that’s how we do it we test we get the feedback we get the feedback until the text off when it takes off it takes off big and if it doesn’t take off it doesn’t matter i use capacity maybe four one two three four five six weeks six weeks 10 servers hundred dollars I lost 100 dollar big deal and actually you learn from your mistakes anyway that’s what we are finding as well so a prototype shouldn’t require you to build data centers obviously you know one of the key concepts here under you know maybe I would say the takeaway for me here for for you here in this session is that you know think of different concepts when you develop ideas at amazon we do something that is called a minimum viable product and it’s effectively the minimum idea that validates my idea and effectively you know let’s put that in the context of digital let’s say I want to create a text editor right I want to provide a new text editor to my customers what will I do will actually write a microsoft word that contains ten thousand features and effectively I only use the bold italic and highlight basically i’m using three of the features of the of the of the suite or will i create a notepad you will create a notepad as what MVP means but the next feature will be driven by what your customers are telling you you build only what customers want that’s what amazon does i’m putting that in the context of Hama Hama some did it so the sexy page here that you see on screen is basically one of the first releases of amazon.com that’s how we started so this was called Oh Beatles actually that was a code name the frame with the Beatles or bidders just for the story is basically the village that is located at the Narrows point of the Amazon on the Amazon River it was actually so narrow that he nearly killed us this application was a big monolith a very very ugly code with a very very ugly code and effectively you know as you might know way if you’re in the world of coding is just trying to plan things on top and on top and on top and on top and basically you end up with something that you cannot get this very rigid and cannot be modified you cannot actually innovate on it so effectively what we did is basically we took a Beatles and we actually started to break it down into different pieces very very small pieces this piece is in the techie well they are called microservices microservices the concept of micro services to say I’m going to have a team a very small team we call that the cube the chip it’s a team it’s a team that if they eat mode that Jupiter’s already too big small teams extremely focused on doing one thing and doing it extremely well basically it’s a mini startup many startups and you give them full ownership on how they want to develop that on how they want to develop that micro service so effectively what it means you give them ownership they are responsible from the conception back to the actual implementation and the other thing they are interesting about it is that and think about microservices you give the team complete freedom when it comes to actually choosing their tools it could be done at Java or whatever and that’s basically what you see on screen here every single feature here so microservices is a team in the back it’s a team running that stuff the main advantage is that when you have microservices you’ve got the cloud all these methodologies that Darlene a gel

DevOps is a marriage in heaven right because you give them ownership and you give the freedom at the same time the only other advantage is that effectively you know if something blows up in one of the microservices the blast radius is really limited it doesn’t impact the rest while we are seeing as while operating this way is that the speed of iteration beats the quality of iteration the more you test the more likely that you have to catch up bugs in advance and we’ve seen that x and x away and effectively what we are we’re at situation rather these are all numbers that we know on the website i believe that the latest numbers is that we can actually update the website every second every second there isn’t an update on the website so one of the key questions you know when you get back to you come to your company I’d actually ask you know you’re actually depend on how long does it take you to deploy your code change the reality that is the wrong question the right question is you know how long it takes you to innovate and the final point because I think I’m running out of time is focused on your business few examples here of customers that are using micro services in connection with AWS nike+ when they talk about you know their experience of the cloud and the cloud using microservices effectively they talk about disaster indifference because in the cloud by default you get dr they talk about loading difference because effectively the elastic concept you need more servers you will have them talk about surgeon difference when you change something on the website if you use microsoft is effectively you know effectively continuous integrals and a continuous deployment technologies effectively it’s completely indifferent to you another one I want you to highlight here is my last slide I’m talking with the lot of retailers talking about what the store of the future looks like you know what how can I bring innovation within my store so they are things such as you know the smart fitting rooms really hot at the moment now we are trying to see you know what how can can we can integrate that with Alexa you know our voice recognition system so effectively you’re in front of view of your mirror and you can put a dress on and you can actually ask the mirror and say mirror change the color of my dress make the list sleep a bit longer mirror take a picture of myself post it in to pinterest and make a plus and a minus to all my friends so that they vote about what they want what I should be wearing mirror you know tell me put this one at this side with in my basket and use these credit card details but they’d other things that we are looking at right now and this is being implemented as we speak you know the concept of having a mobile phone that I do identifies you and actually generate predictions and recommendations based on you know story call data or machine learning mechanisms it’s already done the beauty about these architectures is actually this part how many servers are as an organization you need to manage zero focus on your business you are not managing any service you are interacting with services web services it’s a configuration exercise you do not have to have a guy in the back room that is actually plugging cables you do not have to wait for a truck you want 10 servers you have them you want 10,000 service you have them but still you are talking to an epi focus on your business happy transformation I will be staying here for more question because I think I’m already late more than happy to you know continue this conversation and I hope you enjoyed it thank you very much you