 to the AWS cloud with the AWS storage gateway. Okay, we're back. This is Dave Vellante with Wikibon.org and this is theCUBE, Silicon Angles flagship production. We're here at AWS Summit in Moscone. I'm here with Jeff Frick, my co-host for the week. Jeff, it's really a pleasure working with you. West Coast guy with a tie on. I'd love to see it. I got the tie on for Dave. You know, my colleague John Furrier usually goes no tie, but so we're here with David Ward, Dave Ward, senior manager of EC2 at AWS. We're going to talk pricing. Amazon is really, as we've heard all week, we heard last year at Reinvent, Amazon is very aggressive and proactive about dropping its prices, passing that out to the customer, but it's also innovative in the way it packages pricing. So Dave, welcome to theCUBE. Thank you. So let's talk about the different options. We were just talking off camera about how, you know, one of our colleagues is moving into AWS and you had suggested maybe helping us out with some pricing models. So what are my options as a customer? I'm new to AWS. What should I be looking for? So in general, AWS has a pricing philosophy, which is you want to pay for what you use. As you grow bigger, we want you to have reduced prices. And if you're able to commit, you can get even reduced prices. So that breaks into three main pricing models for EC2. The first is called on-demand. And on-demand really allows you to pay by the hour. So effectively you can try out new instance types. You can try out a bunch of new different services throughout EC2 and only have a commitment of one hour. The second is reserved where you pay a low one-time fee and then a much lower hourly fee associated with any of your usage. And then the final pricing model is spot where you can bid on unused capacity to leverage tens of thousands of machines or and also get pricing savings upwards of 90, 92% off the on-demand price. So help me understand when I should use each. Sounds like on-demand for new. You want to do some experimentation, reserve pricing. You start to get some more predictability in your business model, pay a little more upfront and then spot is just opportunistic. Is that right? Talk about that a little bit. Sure. So typically what you see is with on-demand pricing, it's people who are relatively new to AWS or who are trying to benchmark new applications or really have a lot of volatility in their pricing. Typically they're running between zero and 15% utilization over a one or three year period. When you get to the reserved instances, a customer at that point has made a decision to basically run on AWS and with that workload, somewhere between 15 and 100% utilization over the entire term and they don't want their instances going away at all. They know that they have a certain load that they care about. So think of your web application servers, think of your database servers, anything where you really want to make sure that it's around. When you get to spot instances, we really are talking about opportunistic computing and what that means is that you typically have a couple of different use cases. Two of the most common ones are where you have a certain baseline workload which you're running on reserved instances and you want to accelerate that workload by opportunistically adding more capacity, i.e. spot. Alternatively, you have a certain need-by date when you need to get your workload done by and it doesn't matter the value of your data doesn't really change significantly before that need-by date and so you're able to delay it and so that's another great time for spot. Pretty much most of the other opportunities are on reserved. Dave, you were talking about, there's just one more follow-up that I know Jeff wants to jump in. You were talking about utilization, a range, a spectrum. How are you measuring utilization? Can you just clarify that a little bit? Sure, in terms of utilization, what we want to look at is that one instance running 744 hours costs very differently for us than 744 instances running during one hour. So when it comes down to it, utilization for us means that you're running one instance every hour during the month and that would be 100% utilization. So 30% utilization would be effectively 30% of 744. Got it, okay. So I'm just curious, I thought I understood but now I'm a little confused. What's it, so on demand is I need horsepower and spot I need more horsepower. What's the difference between the two? So the best way of thinking of it is that spot you basically go in there and bid for our unused capacity. So you set your bid price, the maximum you're willing to pay and then we have a spot price which is based on supply and demand. Since we're selling our unused capacity, that spot price will change over time. And if the spot price is below the bid price, you get capacity, if it goes above, you lose that capacity. I don't get it, okay. So we'll actually terminate your instances at that point. Okay. On demand, if you acquire it, right, then I got it. Right, it spins up, it goes. So on the spot, am I buying unused capacity from Amazon or am I buying unused capacity from Dave because he's not using it on his project that he's got reserve stuff? Unused capacity from Amazon. From Amazon, okay. Okay, so I can't arbitrage my unused capacity. I thought it was a secondary market like the ticket brokers. So you talk about that, we actually have something called the reserved instance marketplace. Right. And the reserved instance marketplace is where we have reserved instances that customers have purchased from Amazon and they can buy and sell those. And so we talk about a secondary market. We actually have one of those too. So you do have one, okay. Yes. And that's the reason why that came into existence is that customers were asking for more flexibility with reserved instances. And so at the end of the day, what they were really asking for was a way of saying my business suite has changed. I want to sell this XS1 or get rid of it somehow and then buy another one at another store. So Dave, can you share with us, even in general terms, any data that you have from a customer perspective on how much money I can save in each of these models at scale? Yeah. So let's just start out with on demand. So depending on the best way of thinking about your cost savings and on demand is what is the peak amount of capacity that you would have had to purchase on premise? And if you were to run that the whole time, that's what you would have to provision to meet your potential needs in terms of demand. Versus only running that for one hour, right? And so the gamut there ranges significantly. So it's hard to throw around exact quotes on that one. Unreserved, if we benchmark that against on demand, you can get upwards of 72% off beyond demand price. And for spot, typically on average, we see at least 80% savings. However, we've seen upwards of 92% off beyond demand price. Okay, and then you mentioned the reserve instance marketplace, when did you guys launch that service? So we launched that, I believe it was, July or August of last year? Last year, right, okay. And then that transaction as well goes through Amazon of course, right? So I get the same sort of interface, pricing, transparency, everything else, it's not some new interface. What kind of traction is that getting? It's actually working out really well. It's beating our internal goals. It's in terms of the buyer side, we're seeing a lot of great selection out there. So you can find new pricing terms and great selection of pricing. On the seller side, we're actually seeing, assuming that you price at a very competitive rate, we're seeing that reserved instances are rotating off that marketplace at a mid-term. Is this stuff patentable? I mean, so what's patentable? The methodology that you use to determine all this marketplace and transactions or is that the concept itself or both? So there's the whole gambit of utility patents versus business methodology patents aren't really as patentable there. Yeah, they're kind of fuzzy, right? Yeah, so a lot of this comes down to the utility type patents of how does spot work, how does the reserved instance marketplace work, how does the interchange. And the details behind it to make sure it's accurate and fair. And so you guys obviously have filed and will continue to file patents in that area. It's actually a great area out there for us. There's a lot of white space for innovation. And so it's a great place to be in terms of continuing to innovate and move the field forward. And so we hope to continue to kind of push that trend. Are you hiring people? Oh, I mean, everybody's hiring, right? What are you looking for? So on my teams, I'm looking for a great senior manager, someone that leveling to Amazon is a little weird. We have people that I've had a senior vice president of a thousand plus person company working for me before. That's, so senior directors and directors like at most other companies, that's the level of person that I'm looking for to run the spots engineering work. I'm looking for principal engineers. So people that are super experienced that are really excited about doing things in highly distributed, highly scalable computing, senior engineers, web developers, a product manager as well. So someone that's really, that has both a business and technology background. So... You're exploding. We are hiring all over the place and I would love to connect with anyone who's interested. Aziz, you guys announced, I guess it was in March that you were giving away a trusted advisor, I think for a month to let people try it out. How much does trusted advisor get involved in recommending pricing? That's a great question. We've already made over 22 million dollars in recommendations on how to optimize cost. So people could save 22 million dollars worth of money just optimizing based on reserved instances. So it's one of those things where trusted advisor is a great tool and we're going to continue to innovate in that direction of helping customers to save money. So how's trusted advisor work? Like I say, you sort of did a freebie in March if I recall. And then now it's a service that I can purchase, right? Correct. So the economic justification of purchasing that service is you're going to tell me how to save money and how to improve best practice. So it actually comes along free with premium support and so I believe it's enterprise and I forget maybe it's developer support and along with that you get the trusted advisor service which basically will perform a set of checks against the service you're running and your resources and funds. And what that comes down to is are your resources idle and you're optimizing your service on your services and you, any number of these checks are your security groups set up right and all of our best practices we've been folding into the trusted advisor tool. And so our hope is to continue to expand that over time and you can check it. Now premium support, you mentioned premium support. I presume you're involved in that pricing as well or not necessarily. Premium support not a problem. No, okay. So I'm not even going to ask you that then because I had sort of a detailed question there but it's not fair if that's not your area. How about the Amazon TCO calculator? Do you have any visibility on that? I do have visibility. So that was interesting. I've been playing around with that quite a bit in the last several months and I have to say I give you props because it's an honest calculator. And the reason I know it's an honest calculator is because you actually can make for instance you can make on-premise servers more expensive. I think your storage pricing maybe a little aggressive, a little high for the on-premise so I think I could probably get some on the spot market open but in general I found that calculator to be and I would consider an honest calculator. It wasn't rigged. You know a lot of these TCO calculators are just kind of phony. But that's a service that you put forth to people. Do people, are they actively using it? I mean I know it's about lead gen and getting people to engage but do people actually use that to make business cases? Definitely. So a lot of what we're trying to do the economics are there for AWS to be a, to win against on-prem almost in every scenario. So our goal is really just to provide the data to help people sell from within companies themselves. And so people are using the TCO calculators. People are really starting to build those use cases inside. In fact I've seen some people leverage tools like that. They say this is broken over the weekend they moved to AWS and try something out whether it's there's a large enterprise, an airplane company that was running on us that their testing app was broken. So over the weekend someone did a big spot trial and like all of a sudden they had a testing service that was working better than their provider and the next thing you know like there's a enterprise salesperson going into the lobby and they're like, shh, we haven't told anyone yet. So this is classic Amazon. It's like self-service pre-sales. Yes. Yeah, yeah. And then as shadow IT. It works perfectly because by that time you really develop that trust that hey we are going to really reduce the price. And it was so low of a cost that they didn't even occur to them that even anything was changed. Right. So Dave let me shift gears just a little bit because we had a lot of people talking about kind of the culture. It's a little different. I grew up in Portland, right? So Seattle is Boeing, right? Boeing and Microsoft and Starbucks and more and more Amazon, right? So talk a little bit about what makes the culture special and why you guys are able to continue to execute in you know kind of extensions, extensions, extensions, extensions into new areas. Yeah. So I think a lot of the difference in our culture comes down to we have the decision-making processes really pushed down to each of the teams. So I'm effectively a many general manager. So I can make many different decisions across a small market, across reserved instances and a number of other areas. And what that means is that we're able to operate very fast and very in an agile fashion. And so we have a lot of these teams moving in parallel and any of the dependencies between teams tend to be, we try to cut those whenever possible so that people can move independently. Additionally, there's a huge culture around dive and deep. And so what that means is that as opposed to PowerPoint, PowerPoint's virtually been outlawed at Amazon. That's a good thing. So instead we write Word documents that are these six-page single-spreads papers that are called narratives. Paragraphs. It makes you sit down to the details, right? Yeah, yeah. And when you start to explain things in that way, it means that your ideas actually have to happen. Yeah. We also have a culture of what we call post-holing. And post-holing is this idea, have you ever climbed a mountain or anything of that nature? Yeah. So you actually- More the ice axe usually than the post-hol digger. So the post-holing lets you at least know if there's some soft spots out there. Right, right, okay. And so when we're reviewing these narratives, we can figure out are there soft spots? Have we really thought through the whole idea? And everyone sits around a table reading the narratives and just read them during that meeting and let me talk about it. And it's a very open culture where I've had documentation developers challenge VP before and say, hey, I don't believe this is the right thing. And they had- The authority to really take that through. Awesome. Excellent. All right, Dave, well, thanks very much for coming on theCUBE. It's an interesting story. We really appreciate the insights, the pricing transparency, the speed at which you guys continue to push on price to clients. So congratulations on all the progress you made. Good luck on all those hires that you have to make. We're doing our part to spread the word. So- I really appreciate it. Really appreciate it coming on theCUBE. All right, thanks for watching everybody. We'll be right back. Sajay Krishnan is up. He's the general manager of the AWS Marketplace. You want to hear this story. It's very interesting. So keep it right there. We'll be right back with theCUBE, Silicon Angle Live at AWS Summit in Moscone. Right back. We good? We good? Yeah. Thanks so much. Thank you. Appreciate it. Audrey, let us know when we're out, okay? No, it's okay. It's fine. We're finding the right song. Sajay, Dave Vellante. Dave, Sajay, how are you? Sajay, pleasure to meet you. Likewise.