 Live from Las Vegas, it's theCUBE, covering AWS re-invent 2017 presented by AWS, Intel and our ecosystem of partners. Hi, I'm Stu Miniman, I'm here with my co-host Keith Townsend and you're watching theCUBE's live coverage of AWS re-invent 2017 here in the heart of the Sands Convention Center in Las Vegas. 43,000 people in attendance, spread out across many of the facilities here in Vegas. So, lots of lines, lots of things going on. Happy to welcome back to the program Andrew Hillier, who is the CTO and co-founder of Dentify. Great to see you, how have you been? Good to be back, been great, been loving the show, it's a huge show. All right, so we were pretty excited because we've got a double set here at Amazon's show for the first time, it's our fifth year doing this show. There's another show that we interviewed at where we've had a double set for a few years and that of course is VMworld. We've been watching kind of this change as AWS says, from the old guard to the cloud native, if you will. Talk to us a little bit about what Dentify's doing, how you fit into the ecosystem here. Sure, yeah, and it's a very different show than I think the one two months ago, also right here I think, even this morning the keynote, they've referred to that as what you've been doing the last 10 years and this is all very forward-looking, much more vision and what we're finding is that a lot of the challenges of the past are coming right back again. When you start moving to this new operational model, how do you optimize it, how do you save money, how do you keep pace with change? So today, the keynote was a very different thing than you would have seen two months ago. It's all about innovation, innovation, innovation, not just new feature, new feature. Yeah, Andrew, I want you to talk about the customers that you talk to, the mindset. Are there VMware customers and AWS customers? How are they approaching things like innovation and strategy? Well, I think everybody's kind of caught in between and people talk about hybrid a lot and what we find is that a lot of people's mentality is there's really the cloud and then all that other stuff and one is just one that I'm divest or get rid of and one is really where the mind share is. So even though people might have 10 times as much in VMs as they do in the cloud, they're just thinking about the cloud, that's what they, and then there's a lot of questions about how do I move there, how do I run there, how do I get that bill down because the bills are very, very high when you start running there and there's constantly new services like you saw this morning that also can help make that bill even higher. So how do you get there in a safe way, safety and efficiency? Um, that's why I focus on that cloud optimization piece. You know, we used to talk about, you know, VMware first started, it was like, oh, great utilization efficiencies to be able to kind of consolidate, but, you know, we had VM sprawl and now we, you know, there's cloud sprawl and containers, people are, you know, trying to figure out how things, what are some of the key challenges that you see from customers? Where are some of the big places that, you know, that they can really save a lot of money? Sure, yeah, it's, you know, in the virtual world is all about what we would call playing Tetris where you look at the workload patterns, we do a lot of workload pattern analysis and say that's busy in the morning, that's busy at night, you put them on the same host, it's cheaper and runs better and you can really get huge efficiency gains in those environments. As you move to the cloud, it starts to look a bit different. I'm buying small, medium, and large. So we do the same pattern analysis but say, yeah, that's, you're on an M3, you should be on a T2, you're on the wrong thing and we're seeing around 40% savings on average just by pointing and doing that. So we're seeing massive, you know, it's a different kind of opportunity but it's this equal in magnitude. The savings we saw one customer last week was 57% savings. By just getting the cloud watch data, analyzing the patterns and saying, you're buying the wrong stuff. Now, where this is all going interestingly is that when you start to move the containers, that game of Tetris comes back into play again. You need some much more advanced analytics to say, yeah, how do I combine my workloads to make them fit on the smallest footprint? You know, hence, densify, that's what we do and we're seeing savings upwards of 80% when you do that. So there's huge savings with the right analytics. So with the analytics, much different conversation, as you mentioned, than it was a few years ago to now. You would go to a data center manager and say, you know what, I can really save you on infrastructure costs by optimizing your efficiency. You know what, sunk costs, I don't care that, you know, if I'm oversubscribed, not a big deal. In the cloud, that is a tangible thing. Someone has to pay that OPEX bill. But with that said, even with the optimization, sometimes going to reverse, especially with all the announcements today, you got to figure out, you know what, am I optimal in the cloud, or can I use some of these older assets in my data center, move workloads back there? Do you guys help with that decision matrix of saying, you know what, do I run what's existing in the cloud that's not elastic back in the data center? Yeah, you can do that, what if analysis, absolutely. People do ask that question. So, and again, it comes down to over-committed. If I can actually take multiple workloads and stack them up, some workloads, that's much cheaper than others. And it really depends on the workload. So if you're running in the cloud, we can say those are good where they are, you know, we're working on reports to say that's better off in a Docker container in the cloud. Or if you say, what if I put them back on prem? You can do that too. Now that's kind of regressive. We don't see a lot of people, they're curious about that number, but we don't see people moving backwards. But in the cloud, there's so many different ways you can run the workloads. And they're wildly different cost structures. Again, if you have very peaky workloads that are just busy for a short period of time, to put them in a standard large instance is very expensive. So let's talk about that cloud Tetris in the cloud. Because with containers, obviously, I can now over-subscribe a bit, because I couldn't do that before. You know, if I had an EC2 instance, an M2 large, if I was too big, I'm just too big, and I can't over-subscribe that, and I pay for what I use as advantages and disadvantages. How does that impact the conversations, the design conversations customers are having over stuff like we heard EKS this morning, PowerGate, and all these container orchestration management tools? How does that complicate the conversation? Yeah, it's interesting because if you have a workload that's not doing very much, but you can't turn it off, it's doing something, and then peaks once a day, and you put it in a large M4 or whatever, you're going to pay for the area, you're going to pay for it all, even though you're only doing a little bit of work. So that can be very expensive. That same type of workload, if you put it in a container, with other ones that have a similar pattern, but at different times, they all stack and co-habitate nicely. So that's where we see a huge opportunity to run much higher density and lower cost. But the big challenge we're seeing that affects all the technologies that interest us this morning is that, from a development group, they don't know how to say what those containers need. Right. So we're seeing a lot of Kubernetes environments, a lot of ECS environments that are running very low utilization, because the developers are asking for two CPUs for each container. And it scales based on the number of CPUs or amount of memory, the resource, not the utilization. So we're finding, it's like history repeating itself. I've got the scale group running in containers, very low utilization, and come in and say, well, wait a minute now, that's all wrong. If you do this, and you recommend that, and our analytics give you a bunch of very prescriptive actions, you're going to run much higher utilization. Your bill goes way down again. So it's the same lack of visibility, lack of analytics to figure out how to optimize that equation. Andrew, one of the biggest challenges coming to a show like this is things change so fast. Last year or two, heard a lot of grumbling from customers about, oh, reserved instances, kind of locked me in, it was inflexible, Google was better, Amazon changed what they're doing. This year, a lot of new things on spot instances. The spot market's been around for years, but didn't seem to have a lot of utilization. Amazon's like, no, this is going to be it. If you don't need to have it now, we're going to save you 90% if you do that. Do you help with that? What are you seeing and hearing from customers and how do they take advantage and don't get locked into some huge bill? Yeah, absolutely. I mean, in general, what we find is the pace of change is fantastic for everybody if you had to figure it out. So what we find in customers is that just keeping track of, we have a lot of customers that haven't seen this keynote this morning. They may not be aware of all this stuff yet. So, and even if you are aware, do you know how it impacts you? Can I actually leverage that M5? How does that affect me? So that's one of the things that strength of us is that we deliver a service, not a product, not a tool. So it's analytics, sass hosted analytics, very powerful analytics with a densification advisor, a human that comes with it. So we're on top of these things. So when new stuff comes out, for example, we're sending a message out to our customers right now saying, we're on it, there's three new instance types that came out this morning. Analyzing your environments to tell you if you can use them. So what we're seeing is a kind of a conversion to say that customers, they can't figure it all out anymore. It used to be in the old world, I could buy gear once every three years and I could understand that gear and I could understand my apps. Now you have to pick. Do you want to follow all the news in the cloud or do you want to work on your apps? And what we do is say, work on your business services, your differentiation. We will tell you how it maps to whatever Amazon is selling today. And don't worry about that. And our advisors just do that for you. Yeah, Andrew, I laugh. I think how much of my career was it like, oh well, we're managing that on a spreadsheet and we take care of it. No, no, no, forget about it. I'm curious, the bare metal instance is one that's been a little bit of buzz. It's what they designed for the VMware on AWS environment and offered it, but it's a big honking machine. It's super expensive, but I have to think if you're working with them, you could probably help customers optimize, get great utilization. What do you see that being used for? Is that something that you think your customers are going to be interested in? Absolutely, I think it goes back to that plain Tetris discussion. So if you take one of these big bare metal nodes and run Docker containers, stack property, we see that in one study we did is 82% cheaper than if you put them in small, medium, and large instances. Wow. Again, it's because of the shapes and patterns of the workloads and getting the dovetail. So there is a place for these big monstrous machines like the X1, 32 extra larges, all those things. If you use them right, you can save a ton of money because you get economies of scale. So bare metal is great. It's just another way to host things that for certain apps, certain workloads, makes a lot of sense. For other ones, it makes no sense. So we're at a conference full of developers. They don't care about infrastructure, more or less. And the Tetrix works well when we're talking about containers, EC2-sized instances, even bare metal, what about concepts such as serverless and in which we're just running code? Obviously, we can't make every application based on microservices and it's not practical to take Lambda and build an entire stack. However, there's obviously some opportunity to offer some really incredible savings if we choose Lambda for certain functions. How do you guys help customers make that determination? So, I mean, Lambda is very interesting because there is a break-even point. So if I'm charged every 100 milliseconds for what I run, if it doesn't happen very often, that's a much better way than running an instance. Now, once it gets beyond a certain point, it might be cheaper to actually just run it on an instance. If you have constant workload that's taking up many servers with a capacity, there's a break-even point there where it'll become more expensive to run that because, again, you're paying by 100 milliseconds by the resources that you're being allocated. So if you can run that workload with other workloads and get a common scale, it might be cheaper. So it's kind of, we picture Lambda as almost like the area under the curve. What work am I doing in my app or my service? And if I turn that into how many, and we use benchmarks to normalize everything so we understand that that running there is the equivalent of that running over here. So that workload would incur this many time slices and cost you this much. And so that's something we're working on. It's not released yet, it's kind of coming, but we see that as being able to analyze the equivalency between different models. That workload, very expensive in Lambda, that other one, perfect candidate. Andrew, why don't you bring us inside a little bit? What it's like to be a partner in the AWS ecosystem. Any announcements that surprise you, you guys get some preview on this, how fast can you kind of ramp your team up to take advantage of the umpteen billion new announcements, 1,300 announcements a year. So take us inside how that works for you and then how you help your customers to take advantage of that. Sure, yeah, so I mean, we stay pretty plugged in. We are partnered with all major providers and we do of course, and a lot of the Amazon provides an API to get all the latest and greatest stuff. So if you're constantly hitting that, you get all the latest stuff anyway. So that pace is just built into what we do. We had a customer that said, hey, you guys have really kind of, while we were evaluating your software, you did a lot of things to kind of show the latest stuff coming out of the cloud. Is that just you selling us and you'll stop? And we said, no, that's just the way we operate. That is the new model that everything is just up to date all the time. Like it's, you know, customers are long term, you are getting new stuff every day, we're sending out the notes. So we try to stay on top and again, the key here is that interpreting what that means for you. To know that there's a new M5 is one thing. To know that it's got a different hypervisor and you might need to rebuild your AMIs to run on it. It's something we can say, okay, we're going to help you with this and help interpret this delta. And I think it's a very important thing that, you know, we don't, you know, Amazon has so many priorities of all the breadth and ML also they're doing. They're not really focused on helping you shave your build down. They're just not doing that today. And so I think this feels a very important role in the ecosystem. Andrew, want to give you the final word? You know, what are the things that, you know, your customers are doing that really helping to transform their businesses? Kind of the biggest kind of challenges and opportunities that they're seeing. Well, I mean, I think, clearly, I think containers are going to be a very big part. I know we talked with them a lot already, but I think that's one of the most exciting areas. I think everybody's moving into the cloud and starting to leverage it and doing various ways. But I think everybody's goal, and really Kubernetes is a big part of that. You know, we saw different options out there, really centered on Kubernetes, everybody's doing that. And I think that's really going to be the transformative technology. You can host, you know, born in the cloud microservices, you can host legacy apps in it. You can get to a really high level of efficiency all in the cloud with that one technology. I think it's a real game changer. It just needs to roll through these environments. All right. Andrew Hill, you're always a pleasure to catch up. Thank you for giving us all the updates on Dentify. For Keith Townsend, I'm Stu Miniman. You're watching theCUBE.