(upbeat music) >> Hello, everyone and welcome back to fabulous Las Vegas, Nevada, where we are here on the show floor at AWS re:Invent. We are theCUBE. I am Savannah Peterson, joined with John Furrier. John, afternoon, day two, we are in full swing. >> Yes. >> What's got you most excited? >> Just got lunch, got the food kicking in. No, we don't get coffee. (Savannah laughing) >> Way to bring the hype there, John. >> No, there's so many people here just in Amazon. We're back to 2019 levels of crowd. The interest levels are high. Next gen, cloud security, big part of the keynote. This next segment, I am super excited about. CUBE Alumni, going back to 2013, 10 years ago he was on theCUBE. Now, 10 years later we're at re:Invent, looking forward to this guest and it's about security, great topic. >> I don't want to delay us anymore, please welcome Mark. Mark, thank you so much for being here with us. Massive day for you and the team. I know you oversee three different units at Amazon, Inspector, Detective, and the most recently announced, Security Lake. Tell us about Amazon Security Lake. >> Well, thanks Savannah. Thanks John for having me. Well, Security Lake has been in the works for a little bit of time and it got announced today at the keynote as you heard from Adam. We're super excited because there's a couple components that are really unique and valuable to our customers within Security Lake. First and foremost, the foundation of Security Lake is an open source project we call OCFS, Open Cybersecurity Framework Schema. And what that allows is us to work with the vendor community at large in the security space and develop a language where we can all communicate around security data. And that's the language that we put into Security Data Lake. We have 60 vendors participating in developing that language and partnering within Security Lake. But it's a communal lake where customers can bring all of their security data in one place, whether it's generated in AWS, they're on-prem, or SaaS offerings or other clouds, all in one location in a language that allows analytics to take advantage of that analytics and give better outcomes for our customers. >> So Adams Selipsky big keynote, he spent all the bulk of his time on data and security. Obviously they go well together, we've talked about this in the past on theCUBE. Data is part of security, but this security's a little bit different in the sense that the global footprint of AWS makes it uniquely positioned to manage some security threats, EKS protection, a very interesting announcement, runtime layer, but looking inside and outside the containers, probably gives extra telemetry on some of those supply chains vulnerabilities. This is actually a very nuanced point. You got Guard Duty kind of taking its role. What does it mean for customers 'cause there's a lot of things in this announcement that he didn't have time to go into detail. Unpack all the specifics around what the security announcement means for customers. >> Yeah, so we announced four items in Adam's keynote today within my team. So I'll start with Guard Duty for EKS runtime. It's complimenting our existing capabilities for EKS support. So today Inspector does vulnerability assessment on EKS or container images in general. Guard Duty does detections of EKS workloads based on log data. Detective does investigation and analysis based on that log data as well. With the announcement today, we go inside the container workloads. We have more telemetry, more fine grain telemetry and ultimately we can provide better detections for our customers to analyze risks within their container workload. So we're super excited about that one. Additionally, we announced Inspector for Lambda. So Inspector, we released last year at re:Invent and we focused mostly on EKS container workloads and EC2 workloads. Single click automatically assess your environment, start generating assessments around vulnerabilities. We've added Lambda to that capability for our customers. The third announcement we made was Macy sampling. So Macy has been around for a while in delivering a lot of value for customers providing information around their sensitive data within S3 buckets. What we found is many customers want to go and characterize all of the data in their buckets, but some just want to know is there any sensitive data in my bucket? And the sampling feature allows the customer to find out their sensitive data in the bucket, but we don't have to go through and do all of the analysis to tell you exactly what's in there. >> Unstructured and structured data. Any data? >> Correct, yeah. >> And the fourth? >> The fourth, Security Data Lake? (John and Savannah laughing) Yes. >> Okay, ocean theme. data lake. >> Very complimentary to all of our services, but the unique value in the data lake is that we put the information in the customer's control. It's in their S3 bucket, they get to decide who gets access to it. We've heard from customers over the years that really have two options around gathering large scale data for security analysis. One is we roll our own and we're security engineers, we're not data engineers. It's really hard for them to build these distributed systems at scale. The second one is we can pick a vendor or a partner, but we're locked in and it's in their schemer and their format and we're there for a long period of time. With Security Data Lake, they get the best of both worlds. We run the infrastructure at scale for them, put the data in their control and they get to decide what use case, what partner, what tool gives them the most value on top of their data. >> Is that always a good thing to give the customers too much control? 'Cause you know the old expression, you give 'em a knife they play with and they they can cut themselves, I mean. But no, seriously, 'cause what's the provisions around that? Because control was big part of the governance, how do you manage the security? How does the customer worry about, if I have too much control, someone makes a mistake? >> Well, what we finding out today is that many customers have realized that some of their data has been replicated seven times, 10 times, not necessarily maliciously, but because they have multiple vendors that utilize that data to give them different use cases and outcomes. It becomes costly and unwieldy to figure out where all that data is. So by centralizing it, the control is really around who has access to the data. Now, ultimately customers want to make those decisions and we've made it simple to aggregate this data in a single place. They can develop a home region if they want, where all the data flows into one region, they can distribute it globally. >> They're in charge. >> They're in charge. But the controls are mostly in the hands of the data governance person in the company, not the security analyst. >> So I'm really curious, you mentioned there's 60 AWS partner companies that have collaborated on the Security lake. Can you tell us a little bit about the process? How long does it take? Are people self-selecting to contribute to these projects? Are you cherry picking? What does that look like? >> It's a great question. There's three levels of collaboration. One is around the open source project that we announced at Black Hat early in this year called OCSF. And that collaboration is we've asked the vendor community to work with us to build a schema that is universally acceptable to security practitioners, not vendor specific and we've asked. >> Savannah: I'm sorry to interrupt you, but is this a first of its kind? >> There's multiple schemes out there developed by multiple parties. They've been around for multiple years, but they've been built by a single vendor. >> Yeah, that's what I'm drill in on a little bit. It sounds like the first we had this level of collaboration. >> There's been collaborations around them, but in a handful of companies. We've really gone to a broad set of collaborators to really get it right. And they're focused around areas of expertise that they have knowledge in. So the EDR vendors, they're focused around the scheme around EDR. The firewall vendors are focused around that area. Certainly the cloud vendors are in their scope. So that's level one of collaboration and that gets us the level playing field and the language in which we'll communicate. >> Savannah: Which is so important. >> Super foundational. Then the second area is around producers and subscribers. So many companies generate valuable security data from the tools that they run. And we call those producers the publishers and they publish the data into Security Lake within that OCSF format. Some of them are in the form of findings, many of them in the form of raw telemetry. Then the second one is in the subscriber side and those are usually analytic vendors, SIM vendors, XDR vendors that take advantage of the logs in one place and generate analytic driven outcomes on top of that, use cases, if you will, that highlight security risks or issues for customers. >> Savannah: Yeah, cool. >> What's the big customer focus when you start looking at Security Lakes? How do you see that planning out? You said there's a collaboration, love the open source vibe on that piece, what data goes in there? What's sharing? 'Cause a big part of the keynote I heard today was, I heard clean rooms, I've cut my antenna up. I'd love to hear that. That means there's an implied sharing aspect. The security industry's been sharing data for a while. What kind of data's in that lake? Give us an example, take us through. >> Well, this a number of sources within AWS, as customers run their workloads in AWS. We've identified somewhere around 25 sources that will be natively single click into Amazon Security Lake. We were announcing nine of them. They're traditional network logs, BBC flow, cloud trail logs, firewall logs, findings that are generated across AWS, EKS audit logs, RDS data logs. So anything that customers run workloads on will be available in data lake. But that's not limited to AWS. Customers run their environments hybridly, they have SaaS applications, they use other clouds in some instances. So it's open to bring all that data in. Customers can vector it all into this one single location if they decide, we make it pretty simple for them to do that. Again, in the same format where outcomes can be generated quickly and easily. >> Can you use the data lake off on premise or it has to be in an S3 in Amazon Cloud? >> Today it's in S3 in Amazon. If we hear customers looking to do something different, as you guys know, we tend to focus on our customers and what they want us to do, but they've been pretty happy about what we've decided to do in this first iteration. >> So we got a story about Silicon Angle. Obviously the ingestion is a big part of it. The reporters are jumping in, but the 53rd party sources is a pretty big number. Is that coming from the OCSF or is that just in general? Who's involved? >> Yeah, OCSF is the big part of that and we have a list of probably 50 more that want to join in part of this. >> The other big names are there, Cisco, CrowdStrike, Peloton Networks, all the big dogs are in there. >> All big partners of AWS, anyway, so it was an easy conversation and in most cases when we started having the conversation, they were like, "Wow, this has really been needed for a long time." And given our breadth of partners and where we sit from our customers perspective in the center of their cloud journey that they've looked at us and said, "You guys, we applaud you for driving this." >> So Mark, take us through the conversations you're having with the customers at re:Inforce. We saw a lot of meetings happening. It was great to be back face to face. You guys have been doing a lot of customer conversation, security Data Lake came out of that. What was the driving force behind it? What were some of the key concerns? What were the challenges and what's now the opportunity that's different? >> We heard from our customers in general. One, it's too hard for us to get all the data we need in a single place, whether through AWS, the industry in general, it's just too hard. We don't have those resources to data wrangle that data. We don't know how to pick schema. There's multiple ones out there. Tell us how we would do that. So these three challenges came out front and center for every customer. And mostly what they said is our resources are limited and we want to focus those resources on security outcomes and we have security engines. We don't want to focus them on data wrangling and large scale distributed systems. Can you help us solve that problem? And it came out loud and clear from almost every customer conversation we had. And that's where we took the challenge. We said, "Okay, let's build this data layer." And then on top of that we have services like Detective and Guard Duty, we'll take advantage of it as well. But we also have a myriad of ISV third parties that will also sit on top of that data and render out. >> What's interesting, I want to get your reaction. I know we don't have much time left, but I want to get your thoughts. When I see Security Data Lake, which is awesome by the way, love the focus, love how you guys put that together. It makes me realize the big thing in re:Invent this year is this idea of specialized solutions. You got instances for this and that, use cases that require certain kind of performance. You got the data pillars that Adam laid out. Are we going to start seeing more specialized data lakes? I mean, we have a video data lake. Is there going to be a FinTech data lake? Is there going to be, I mean, you got the Great Lakes kind of going on here, what is going on with these lakes? I mean, is that a trend that Amazon sees or customers are aligning to? >> Yeah, we have a couple lakes already. We have a healthcare lake and a financial lake and now we have a security lake. Foundationally we have Lake Formation, which is the tool that anyone can build a lake. And most of our lakes run on top of Lake Foundation, but specialize. And the specialization is in the data aggregation, normalization, enridgement, that is unique for those use cases. And I think you'll see more and more. >> John: So that's a feature, not a bug. >> It's a feature, it's a big feature. The customers have ask for it. >> So they want roll their own specialized, purpose-built data thing, lake? They can do it. >> And customer don't want to combine healthcare information with security information. They have different use cases and segmentation of the information that they care about. So I think you'll see more. Now, I also think that you'll see where there are adjacencies that those lakes will expand into other use cases in some cases too. >> And that's where the right tools comes in, as he was talking about this ETL zero, ETL feature. >> It be like an 80, 20 rule. So if 80% of the data is shared for different use cases, you can see how those lakes would expand to fulfill multiple use cases. >> All right, you think he's ready for the challenge? Look, we were on the same page. >> Okay, we have a new challenge, go ahead. >> So think of it as an Instagram Reel, sort of your hot take, your thought leadership moment, the clip we're going to come back to and reference your brilliance 10 years down the road. I mean, you've been a CUBE veteran, now CUBE alumni for almost 10 years, in just a few weeks it'll be that. What do you think is, and I suspect, I think I might know your answer to this, so feel free to be robust in this. But what do you think is the biggest story, key takeaway from the show this year? >> We're democratizing security data within Security Data Lake for sure. >> Well said, you are our shortest answer so far on theCUBE and I absolutely love and respect that. Mark, it has been a pleasure chatting with you and congratulations, again, on the huge announcement. This is such an exciting day for you all. >> Thank you Savannah, thank you John, pleasure to be here. >> John: Thank you, great to have you. >> We look forward to 10 more years of having you. >> Well, maybe we don't have to wait 10 years. (laughs) >> Well, more years, in another time. >> I have a feeling it'll be a lot of security content this year. >> Yeah, pretty hot theme >> Very hot theme. >> Pretty odd theme for us. >> Of course, re:Inforce will be there this year again, coming up 2023. >> All the res. >> Yep, all the res. >> Love that. >> We look forward to see you there. >> All right, thanks, Mark. >> Speaking of res, you're the reason we are here. Thank you all for tuning in to today's live coverage from AWS re:Invent. We are in Las Vegas, Nevada with John Furrier. My name is Savannah Peterson. We are theCUBE and we are the leading source for high tech coverage. (upbeat music)

>> Live from Las Vegas, it's theCUBE! Covering AWS re:Invent 2018. Brought to you by Amazon Web Services, Intel, and their ecosystem partners. >> Hey, welcome back everyone! It's theCUBE's live coverage, day two, of three days of wall-to-wall coverage. Keynotes, amazing announcements, great vibe here. Again, 52,000 people here at Amazon re:Invent, I'm John Furrier, my co-host Dave Vellante, we're here again, 6th year, just an amazing amount of content, two sets Eron Kelly is the General Manager of the Enterprise Services Marketing Basically part of marketing for Lilac Groups a lot of the areas. Great to have you on theCUBE. Thanks for spending the time! >> It's great to be here John and Dave, I really appreciate it. >> Alright so, how hard is your job? (all men laugh) I mean we were just doing an analysis of Andy's keynote and there's so much to talk about you have so many things under your purview. You have a broad portfolio of Amazon services It's really crazy good for you guys, the business is going great, highly profitable How do you do it all? How do keep track of it? What's your favorite child? Tell us >> It's a great point, there's so much going on and the speed and the pace of innovation but it's exactly what builders are looking for, right? They want to be able to come to AWS and not have to compromise. Because they want to see every tool that they need for every job And so for me I have a pretty broad portfolio I've really been excited this week about a lot of our compute announcements, right? So our announcements around our new instance family with A1 based on some custom silicon AWS Graviton processor Really excited about that Bringing the Arm community to the Cloud for the first time, we're jazzed about that one. >> And that motivation there is lowering costs for Arm-based apps, right? >> It's really two-fold, right? It's bringing the Arm community to the cloud It's the first time we've got an Arm processor in a large Cloud provider, so giving them that scale-up, elasticity, pay-for-what-you-use kind of model but then the second thing is lowering costs for customers for scale-out workloads so things like web tiers containerized microservices and you know in the general purpose area customers, we think they can save up to 45% which is meaningful, so we're really excited about that. That's been a really neat announcement a neat project we've been working on for a while. I would also say in the area of Compute We've added 100GB networking to a couple of our Instance types. C5n based on Intel processors and that's really been the workhorse in the HPC community and now with 100 GB of networking we're going to be able to do even more processing, more power, more advanced scenarios there And it's kind of an interesting dialogue, right? The more data you have, the more compute you need and it creates this virtuous cycle and one of the gaps was networking, so bringing 100GB It really allows those Intel chips to run >> Well, Big Data guys are going to eat that up to, I would think I mean, the links between Big Data and HPC become clearer >> Exactly right. >> Talk about the latency thing Andy talked to me last Monday about this I had a long conversation with him about it Latency matters, you guys listen to customers. Networking, you mentioned a key part of the flywheel Compute storage networking obviously morphing with the Cloud while you guys are optimizing and raising the bar. How are you guys handling the latency question? Because this comes up a lot You got the On-premises piece now you guys are doing things in the Cloud How do you market that service? How do you handle the latency? Talk about the role of latency in the portfolio. >> There's a couple things in there The first one, what I would highlight is Latency within an HPC environment or a machine learning environment and so that's where again this 100GB networking has been really powerful. We've also announced a new networking feature or protocol, Elastic Fabric Adapter which actually allows you to go even faster now in some networking scenarios, which is particularly interesting, again, in HPC So we've really worked hard in reducing the latency throughout the data centers for these higher end compute scenarios. >> Well, you have the custom silicon I wanted to just take you through this because EC2 has always been great, but setting up an Instance could take 10 seconds. Lambda now you can do it in hundreds of milliseconds. By having the custom silicon, How does that impact the network stack? Because I would imagine that the performance gains on having kind of a custom silicon around EC2 and the Compute, would be a gamechanger for running things under the covers, for instance, like a Vmware or managing security boundaries issues, what's the... >> So we made these investments a few years ago with the Nitro system, Where we took a look at the current Instance environment and said hey if we can offload some of that computer networking to a purpose-built chips on those cards we can actually free up more capacity for those processors to run faster and give you more value basically for each Instance type. So that was part of the beginning of Annapurna Labs and the Nitro system was offloading this networking into those custom built chips. That's the start and then what we've done is the Nitro system has allowed us to innovate much faster So we've added three times more Instances this year than last year because we're building on the Nitro system Graviton processor is just one more example We've added new processors from AMD And of course we've continued to advance with our Intel chip set as well. >> Talk about, I just want to just change gears for a bit Because you're in product marketing Executive General Manager Andy talks about the new way the culture at Amazon Old guard, new guard... Traditional product marketing, you can take a product you can bring it to market, waterfall it out at the beach and then you do all the activities How do you raise the bar in your job where you got to go out and take not just products, this is services now so you have a series of, a lot of services How does that change how you do product marketing? And how's that different from people who might not know how you guys operate? Talk a little about the culture of product marketing at Amazon. >> Sure, yeah. So I would say first and foremost it's all about education, right? So we really want to make sure that whenever where there's a new service that comes out we're super clear on what does it do what is the benefit and how can customers take advantage of it. And we're trying to position it in a way We like to say internally, sort of a non-technical CIO can understand So whenever you look at a new service You look at our detail pages. We put a tremendous amount of rigor and clarity We make it very simple, make the value pop clear I think the second thing we're starting to do and we're seeing it reflected in our products as well Is how can we tell more aggregated stories? So today during the Keynote, You saw Andy talk about abstractions, right? And one of the first ones he mentioned was Control Tower, which is one of my products So I got a little passionate around it. But what's interesting about that is that customers are coming to us and saying OK, I love the AWS, I love all these tools I love the granularity of managing things at a certain level and setting policies at a certain level But you guys have thousands, millions of customers running AWS, what's the right way to set up my environment? Can you give me a blueprint to do it? To set it up and run it in a very secure and compliant way. So Control Tower is a great example of a both a service that helps you do that as well as a marketing message that says Hey, let's look at this now in totality Let's you set up these environments faster based on best practices and now you can control in a much easier way. >> So you're basically trying to simplify the message so it's not speeds and feeds >> Well, what I would say is we want to simplify the message so that everyone can understand but we don't want to lose track of those builders those tinkerers that get in there, they want the speeds and feeds, they want the nobs and part of their differentiation as a developer is understanding all the details So we want to have both. >> It's also trying to help people figure out what to use where. As your portfolio grows and grows and grows the complexity becomes amazing for some people And Tailgate helped me figure out mapping to my workload, what to use where what's the best cost solution sometimes it's hard to figure that stuff out, isn't it? >> Yeah, well, it's again, it's this balance between We want to be able to provide the right tool for the right job I think Andy had a nice analogy today in the Keynote Where he talked about building fixing a house with just a hammer, right? And instead you're going to want to have that right tool for that right job And so part of our job in Product Marketing is making it very clear When do I use this particular Instance type versus this versus this? What are the trade-offs? And that's a key part of our job. >> And that resonates with people because there's a lot of redundancy in tools, too in the marketplace, people, a lot of them have the same tool the same hammer and you guys have a variety of services. So the question I got to ask you though As you look at the services and Amazon's role here at the event How would you summarize what's going on here? Because there's so much, Andy had a slide up there that said "Signal from the noise" and that's our phrase Extracting the signal from the noise which is kind of fun, but you have so much signal this noise and there's too much signal how would you encapsulate, for someone watching, what is happening this year? Where is Amazon for customers? What's the positioning? How should they think about, you mention builders? How would you summarize what the action is going on here? >> Right, so I would talk about it like this First and foremost, I would say we're adding more capability and building the broadest and deepest platform so that builders can always, they never have to compromise they always can find the right tool they need for the right job. So first and foremost, that speed of innovation that pace of innovation is continued and if there was one message that people should take away is Wow! They are still innovating like crazy they have an amazing amount of technology and so I don't have to compromise when it comes to datalinks. So that's kind of the first main message. Then I think the second thing I would say which kind of follows on that is OK but we also recognize that we've got a lot of services and now we need to start to build some services that bring these together and again Control Tower is a great example of that Lake Foundation is a great example of that for datalinks. And so that's sort of the second thing we are starting to create services that are abstraction layers that bring together a lot of the details to solve very specific problems. The third thrust that I would highlight is just the amazing focus around machine learning and AI and just how that has been such a key investment area for us and such a key ask for our customers and our mission there is to just democratize it. We want every builder to able to bring machine learning and add it to their solutions. And the number of services you saw announced today in the keynote as well as some earlier this week and last week just shows that our commitment and focus on that >> And extending EC2 to support some of that stuff >> Exactly right >> And the training and the like >> I'm a Star Trek fan so I always go to Scotty Scotty, more power! You guys are bringing more power to the table with each Compute and these abstractions. I want to get your thoughts on something that I talked with Andy about in depth last week before the show, and we were riffing on this notion of a new kind of Developer emerging and he talked about in the keynote, a new persona of developer, A new kind of developer is emerging. And he also kind of talked about net new workloads I wrote about it in my stories on SiliconANGLE in the forums about it, which is this all this goodness going on at the abstraction layer with a lot of horsepower enabling things that were hard to do. AI's a great example AI's been around since I was in college in the 80's and 90's and now it's rose up with power What are some of those persona developers look like? How do you look at that net new workloads? What's your reaction to that? Because this seems to be a big trend that's not your old school developer banging out code now there's OpenSource communities, we got that but in the working day and life of companies people are building apps. What's this new persona developer look like? >> Well, there's different personalities, right? There's the core tinkerer like you talked about there's now the data scientist that's coming in and taking advantage of these machine learning tools. You have kind of a cloud administrator that's kind of trying to look across everything and they want to build as well, right? They don't just want to sit there and manage the dashboards, they want to build as well and so we're seeing that in some of those personas. of course, app developers is another big part of it. >> Now you could talk to Firecracker too, right? >> We could talk a little bit, yeah >> So I met with Adrian yesterday and of course people used to poke at Amazon a lot hey, what about Open-source, you guys giving back to Open-source? And so Firecracker explain that and sort of what you guys are doing there and specifically in Open-source? >> So that's a great example of where we had some technology, and what Firecracker is it's a container for microservices that can run in a non-virtualized environment and we've used it as the underpinning for Lambda and Fargate And we looked at it and we said you know what? We want more people to be able to take advantage of this because it's about saving money it's about improving security. And so we decided to open-source it. And so that was one of our announcements on Monday was open-sourcing Firecracker and making it available to the community and so we're really excited about it. >> One of the things I want to ask you as we wrap up here, first of all great job on all the work you've done at Amazon impressive to see the level of services >> Thank you! That you guys are announcing and it's become a competitive advantage and you've got a great trajectory a lot of learnings and as Andy says there's time compression for experience and time which is good for a competitive strategy but as you look forward to 2019 What's your plan? What are your goals? How are you going to raise the bar? The term you guys use a lot. What's your goals? What are you trying to accomplish? >> You know the number one thing that we spend our time on is listening to customers and saying what's next? What do you need next, right? 90% of our innovation comes from customer input and so now we got a new wave of services we're introducing we're going to spend time with customers they're going to give us feedback we're going to make those services better and then we're going to find new places where they want us to go so next year is going to be just as exciting as this year and next year when I see you guys here we're going to be talking about a whole new wave of things coming out, it's been fun. >> You're certainly running hard and the other thing I've noticed in learning how Amazon works and getting deeper under the covers there you got a growing field, great professional services and a sales force that is not trying to grab the wallet from the customer you guys have a long game perspective >> That's right >> Carla, had a great conversation with her about this you have to service that How are you going to enable these guys You mentioned education earlier this is a big part of your plan, right? The integration with the field how does that work? You going to provide the messaging all the tools...how..cuz that's grow you've got to service that What's your perspective on the field >> Oh for sure, you're highlighting my Q1 goals right now It's really important to dial up that connection because as we get more and more services our field sellers, what's great about our field teams is that they're so aligned to customer needs So they don't carry specific quotas on individual products and things like that they're really focused on hey, what do ya need? And how can use the full portfolio to help you out? And so part of our challenge as a product marketer is not only educating customers on our products but educating the field on our products and which ones are most viable for which scenario and so that's a big part of our focus as well within the product marketing function is hey how do we really nail these scenarios very crisp, very compelling both for the customer and field seller to say ooh I saw that pattern in a customer! Let me go bring this technology forward and talk to them about it, so really excited about next year and you hit on something else that I think is really important which is this long term view, is our sales teams have always taken a long term view with customers they're not sitting there at the end of the Quarter trying to, you know, close the deal it's all about that long term view And it's allowed us to make some of these investments we've had, we've made >> You guys also use your own technology too I noticed a lot of the different groups You've got all the goodness of Cloud. You got to use some of that tech You've probably got some machine learning waiting around the AI bots and all kinds of cool tools you're integrating in dog-fooding or what do they call it? >> I mean, the reason why we're able to add so many services every year is that we build on our own platform, right? So you can have a very small team we talk about the two pizza team A very small team is able to build these services because they use AWS and in its entirety. And so it's very very exciting as these things get connected Like last week we talked about a predictive auto-scaling, so one of the features Auto-scaling's been a pretty popular feature over the years where people can scale up and scale down a large fleet of EC2 Instances but now we've added machine learning to that where it will now predict when the scaling should happen. So it allows you to scale up your EC2 Instances ahead of time based on historical patterns. So there's ML coming into everything we do. >> And server-less is booming too you know, that's going to be a big part of your focus by the way, you mention the Fleets I love this, we haven't talked about it much on theCUBE, but this notion of fleets is pretty powerful Having just a bunch of fleets of servers ready to go >> Ready to go and being able to manage across the different pricing models is also very, very powerful. I want to really ramp up very very quickly take advantage of those spot instances in those moments with really big cost savings as well as ramp back down >> Hey, keep adding functionality keep removing all the barriers lowering the price, making it high performance and I love that business model a lot of companies don't have that so Congratulations! >> You know, we just always want to add more we want to give customers more tools so they can have the right tool for the right job we want to give them the most powerful platform so they can do the highest end things as well as give them scenarios where hey, it's a little bit lower cost and it's for smaller workloads we don't want to overpay and be over-provisioned that's a key part of our strategy. >> I want it all and I want it now >> That's right! >> Well, you did a good job on the messaging Andy wants to mention builders and right tool for the job I think there's a drinking game going on on that he mentioned multiple times Congratulations Eron Kelly, Thank you! >> Thanks John and Dave, Really appreciate your time on theCUBE. >> General Manager Amazon Product Marketing here inside theCUBE, breaking down what's going on, what's his goals how Amazon keeps up with the pace Good insight I'm John Furrier, Dave Vellante Stay with us for more coverage after this short break. (upbeat music)