 from Las Vegas, it's theCUBE, covering AWS re-invent 2018. Brought to you by Amazon Web Services, Intel, and their ecosystem partners. Well good afternoon, welcome back to AWS re-invent here on theCUBE, we are live in Las Vegas, day two of our three days of coverage here on theCUBE, along with Justin Warren, I'm John Walls and we're now joined by Andrew Hillier who's a CTO and co-founder of Densify. Andrew, good afternoon to you. Good afternoon, how you guys doing? Thanks, oh, doing great, how about you and the show? Let's first off, just give me your take on the show and I'm going to jump into Densify, but what's the vibe you got? Well, I mean, it's a great show. It's this show just gets bigger and better every year. I think it's become kind of the de facto industry show. If you're a company doing anything, any startups, any venture capitalists are all here now. All the other shows have kind of, their legacy, this is the one to be at. So it's always really exciting and we always have a lot of travel. We have two booths here and it's just, we're always just flooded with people. So it's a great experience, great event. All right, so Densify, tell us about Densify for those who are watching, maybe not be familiar with what you're up to. So we're optimization analytics. You can think of us as machine learning that learns the patterns of workloads and figures out how to optimize those workloads, the resources you're giving them. So in a public cloud environment like EC2, we'll look at the workload patterns and look at everything that Amazon sells and say, well, how do you align those so you're buying the right things? And it's a pretty big problem. We find that there's not necessarily that level of depth and visibility in a lot of organizations. So they're buying the wrong stuff. So we do the main EC2 services, RDS, things like that, and increasingly containers. Containers are coming up more and more and they're really powerful, but they're also complicated. So the big focus of ours now is on containers. Yeah, it's just been containers wall to wall pretty much in the last couple of days, John, where people are talking about how we want to use containers for lots of workloads, but we're also hearing that it's complex and difficult to manage and that people really, really do need help with that. Or like we've had at least two or three conversations where it's beyond human comprehension. And we need to have essentially augmented intelligence where we need tools that will help the humans to make those decisions. So how does Densify help the humans decide how to use containers? Yeah, and that's a very good way to put it, is at some point you exceed what humans can kind of comprehend. So containers are fairly complex and people think they're kind of magical. I think I'm going to go to containers and everything's going to run great and be agile and they are really powerful. You mean it isn't? Yeah. Well, we're doing a lot of container analysis in our customers and we're seeing that that's not quite the case. And one of the problems is that you still need to say how much resources you need. So when somebody makes a container, they say I need 500 millicores or a whole CPU. And if you get that wrong, the whole thing kind of breaks down because these container orchestrators aren't that smart. You ask for a CPU, I'm giving you a CPU and once I give it all the CPUs I start a new server. And so we see these environments, they're agile, but their utilization is terrible. And maybe even worse than other environments because people just think it's going to be magic and they don't even really worry about it. But it is a big problem. We're seeing in every single environment we've analyzed, they've been really low utilization, these environments. Okay, so we have these environments where they're not magic. So how do you add magic to that? Please tell me that you add some magic. Oh, we do, of course. Everyone loves the magic. The magic is called science and taking opinions out of it. And what we do is we get the detailed workload data and in the container world, that's different. You get more data, it's right down at the container level and you learn the patterns and you figure out, well, how much resources do these things need? And how do they fit together like a puzzle, like a game of Tetris? And so we do that analysis and then we provide much better answers as far as how to specify resources and then the environments start to run magically, they start to run much more efficiently. And so we kind of do that analysis and generate very specific outputs. And the interesting thing is in container environments, in most cloud environments now, you can't just go and call an API and change the container to a different size. Because it'll go back to whatever it was in the Terraform file. Or the cloud, it's coming from usually a DevOps tool chain and that's what deploys it. So even if you know what it should be, you have to actually automate that in a very different way than an old environment. What's the trick or maybe the magic? Because applications don't operate in a static environment, right? I mean, there are a lot of different factors, different variables that come into place. So optimization changes all the time, right? I mean, that's constant. So how do you balance that out in terms of workloads and applications and their performances? Well, it's a great point because it's very complex and if you get any of it wrong, you can't automate something that's wrong. So one of the things that we pride ourselves in is we have a patented rule engine that understands that if you go from that type of instance to that one, you're going to lose enhanced networking or EBS optimized. So we know exactly what can and can't happen on many different dimensions. We know the workload, we know business constraints, technical constraints and then so when you have all that and want to answer, then we can say here's exactly what you should do and it's better than a human could do. If you don't have all those things then you give an answer saying, I think you can do this but you should go and verify it, take two days and figure out whether that answer's right. We don't do that, we give you the right answer. And so when you're talking about automation, that's the hard part, is knowing exactly what to do. The doing it is actually relatively simple but you can't do something that's wrong. People won't turn the keys over the machine if the machine isn't right, it's basics. That is a good point, Andy Jassy in the keynote today went through the laundry list of what AWS does and that it has so much more than any other platform and he was just talking about all of these different options that you can have so that you can pick exactly the right solution for you but because there's so many options, it actually becomes quite difficult for any one human to run through them all and figure out which instance type should I be? Should it be this one? And then Amazon comes out with a new one every six months. We keep changing things. Yep, they came out with new AMD based ones recently. There's all kinds of new ones. So our view is that people should focus on their workloads and their apps and not worry about what the cloud providers are providing. Let the analytics keep up with that. It's a moving target and you're constantly reanalyzing. And so for example last year at this time in this show, they had just announced a new instance type and by that afternoon we had reanalyzed our customers onto it saying, customer X, you could leverage that in 50 or 100 of your instances. And so that's what we try to do is kind of take that complexity out of the equation and allow you to kind of just programmatically say how you want your environment to behave and then we will constantly align it. We call it continuous optimization. So we're constantly watching the workloads and watching what the providers are selling and then giving recommendations of what to do and importantly actually doing it if people are ready for automation, doing it closed loop so that the humans don't have to be involved anymore. Very few customers would be in the business of being the most optimal choosers of AWS infrastructures. Like most of them are just trying to sell products to customers. Why are you spending all of your time doing this? Why not outsource that to someone like Densify? Yeah, absolutely. Your core business is banking or concert tickets or whatever the case is. Just not watching Amazon. You can do that programmatically so people can focus and not be distracted by all these things. People might want to approve changes initially if you're recommending you change something. People on applications might want to be comfortable with it and so we produce reports for application owners as well saying here's what we're recommending, here's why you should be comfortable with it but beyond that they should focus on their business. Coders should be coding, not sizing VMs. That's not their business. So for customers who've got existing workloads that are not containerized and there's plenty of those, I know that Densify's got plenty of experience in the world of virtualization and workloads that might be sitting on VMware or maybe even on bare metal. What is the story that you have for customers who are looking at containers and thinking I would like to move some things there but I'm just not really that comfortable with how to do that. How would Densify help them with that transition from doing a virtualized workload into container land? So with the analytics you can kind of do water analysis in almost any direction. So we take all the workload models and we normalize them using benchmarks. So we can tell if something's running on-prem on a DL380 what will that look like in an Amazon M4? Or an M4 what will it look like in a T2 or a T3? So we can analyze across different platforms across different clouds and so you can do exactly what you're describing. You can say, I have this data center. Recently we took one of our customers and migrated a data center they had in Vegas into Amazon. You say, here's exactly what you should buy in Amazon and what it's going to cost. If you're going into containers we can say, well, if you put them in containers here's how you specify the resources and here's how much infrastructure that will need. So it is an important thing anytime anybody is trying to move to a next gen you want to take the uncertainty out of it. You want to figure out how they're going to get there how they're going to budget for it. So we do do that as well. A lot of our customers use us on everything and then they're constantly shuffling between these operating environments. Virtual into cloud, cloud into containers. It's just kind of a regular, change is the new normal, right? So they're just constantly doing that. So if you were talking right now to a customer who's not there yet, who hasn't made that decision and you're talking about optimization, how do you convince them to let go and that it really are going to give you a better solution to let you focus on your core business whereas they're not used to doing business that way. That's not how our IT works. Yeah, it's a good question because different people have different perspectives. So I could go into you and say, it's going to save 40 or 50% on your cloud build. And that's a pretty compelling. So if you're talking to a CFO, they love that. But if you're talking to an operational team that's not necessarily their problem. They're giving the teams what they're asking for and so they respond more to a automation might be their goal, right? So there's various facets of the benefit. What we find is that the precision of aligning supply and demand so that you're running your cloud property, it helps performance, it helps cost, it helps a whole bunch of things. And depending on who you're talking to, different things will be important to them. So we find that, again, for the most of the people that we tend to work with cloud ops teams and the automation is very important in making their lives easier. So they're not constantly trying to figure this out. They can just let the machine learning do it. Yeah. So what do you think is going to come after containers? What should we be looking at? What have you seen here? What's on the horizon that customers should be looking at that's going to be coming up next? Well, clearly there's serverless. We hear a lot about server like Lambda and things like that. I think that's an interesting technology, but I think it's also not magical. So it's probably really good for certain types of workloads, but not others. I mean, if you use it enough, it might make more sense to put it in a container in your workload. So for transactional workloads, maybe it doesn't make sense. So I think there's constantly more and more diverse offerings and they just kind of abstract things in different ways and the challenge is going to be to figure out which one to use for a given workload. We're already, I was on a call this morning where, what is a good candidate for containerization? How do you identify that? Well, science can identify that. And that's the problem now. And then maybe once you pass that, well, what's a good candidate for, there's always going to be a new technology. We're going to be sitting here five years from now when there'll be something completely different that we've never heard of yet. But it's kind of the same problems in a package in a different form. Well, Andrew, thank you for the time. We appreciate that. Glad this show's going well for you. And we appreciate the opportunity to visit today. Good to see you. Thanks, guys. Good chat with you. Andrew Hillier joined us from Densify, back with more from AWS re-invent right after this.