 Live from Las Vegas. It's theCUBE. Covering VMworld 2017. Brought to you by VMware and its ecosystem partner. I'm Stu Miniman, here with my co-host John Troyer and you're watching theCUBE, SiliconANGLE Media's production of VMworld 2017. We're the worldwide leader in live tech coverage. Happy to welcome to the program, not only a first time guest, but a first time for the company, Andrew Hilliard, who is the CTO and co-founder of Densify. And not only the first time we've identified, we didn't even have Serba on. So I'm not sure what the problem was, but appreciate you joining us and looking forward to learning about you and the company. Glad to be here, it's good. All right, Andrew, tell us a little bit about, you know, you're a co-founder, so you know, bring us back to the early days, what the idea was, and then there were some rebranding recently, so I know that's relevant to the conversation. Sure, I'll tell you the story. So we're all about analytics. I mean, we started off by looking at, you know, all the data that's available and saying if you really do the math on it, you can make a lot of very important decisions and not leave them to opinions or chance. So we built out a very powerful analytics engine. A lot of big customers adopted it, run on-prem, drive huge savings in virtual environments, de-risk, and what we found was that everybody's really interested in those outcomes of the analytics, but not necessarily wanting to adopt software products. I mean, it's kind of the basis of all SaaS. So we went and made a SaaS version of that product that runs like a brain in the cloud to get the same outcomes, and we've kind of really now taken that to the extreme where it's as a service now and it's called densifies. We rebranded around that in the June timeframe to really capture the simplicity and the outcome of what we do, which is to drive down cloud costs, drive down the amount of infrastructure you need on-prem and make it all work better. Yeah, I'm wondering if you can give us just a little from a macro standpoint software and the different consumption models, you just walked there a little bit, but what are customers looking for? Why has it been challenging before? And do we have it right this time? Yeah, well, from our perspective, again, I think we get adopted and traditionally in the past, you would have to deploy the product, so you'd have to provision servers to run it on, a database server, train people, have maybe a set of excellence around using it. And that's worked really well, but I think the novelty of running software has worn off for most organizations. I think they want to move on. As we see cloud being adopted, people just want to get out of the business of running anything, really, and have it all done for them. So we support the on-prem model and as a managed service on-prem, but really this new model is where everybody's going because it's just so simple. It means you can just adopt it and get results right away without reading any manuals or doing anything. Andrew, we've been talking about cloud for years now, right? It was almost a joke, it's much more real now. Your customers and the people you talk with, hybrid cloud, multi-cloud, how many, we have a choice of many different platforms on-prem is not going away anytime soon. At least, I don't know, I love your opinion on that, but your customer base, the people you talk with, what kind of, how many platforms are they on, what kind of platforms, and how does Densify pull all that together? Yeah, it's funny because there's a bit of everything and that's like IT, right? You'll always have one of everything you ever had plus all the new stuff. We, of course, see still huge virtual footprints out there. A lot of companies have big VMware environments, but there's definitely a big focus on the cloud. So almost every customer we have is in some form looking at, it is really that they see that as a future, the cloud, containers, some mix of on and off-prem. So I think it's going to be hybrid for quite some time. I don't think you're going to see the on-prem go away. That would just be unrealistic, but again, a lot of energy is being put into the public cloud and it shows. So one's almost a maintain mode in some cases. One's kind of the invest, we're investing in this new technology is where a lot of the excitement is. So even our most conservative customers are looking at cloud in some way and some of our newer customers are 100% cloud. There's no on-prem. Andrew, talk to us about the relationship with VMware that you've had and have today and I guess one of the questions I looked, VMware announced like seven SaaS services, one which was cost insight. Does that compete at all against what you're doing? Well, it's a hugely complicated space with a lot of different and a lot of the same words used for the same thing. So we have a good relationship with VMware. We integrate with all the product line, VRA, VROPS, DRS, PDRS, we have integrations with all these things and it works with that. But I think there's some confusion sometimes around people using the same words, like we optimize or we do this or that. So what we find is that in the core of what we do is we analyze workload patterns and it's like playing a game of Tetris. It's like saying that workload is busy in the morning, that one's busy at night. We combine them together, we get a lot of efficiency and nothing in the VMware product line does that. So it really plugs in really nicely with DRS and again, VROPS. But there is confusion in the words people use and people might think that does that. And there's some cost, there's a lot of products that do cloud costs. Every product that starts with the word cloud does cloud cost. But that's not really where you get this cost savings. It's really analyzing the workloads in the cloud is where you get the real cost savings. Yeah, I'm curious, you must have a really good view as to utilization. So I think back, there's lots of arguments as to how much utilization are we actually getting? Because VMware in the early days, it was like, oh, I'll consolidate servers, I'll get greater utilization. But we still kind of stink at utilization. When I have gear, even cloud today, I've seen lots of companies write that are, I can take huge amount of costs out of what you're doing. So how are customers doing? What are they good at? What do they suck at? And where are some of the things that you're helping really well? I mean, you struck a nerve there because people are doing a terrible job in the cloud quite often. A lot of times they throw things up there and they don't even really look at what they're doing. It's kind of primitive in terms of the data collection and the tooling around that. So a lot of times people don't even know what the workload is or what the utilization is. So we see some pretty big opportunity to carve that down. I mean, on-prem, I think people have gotten better. I think when they run our product, they really, it's designed to get at the optimal utilization. And that might be 80, 90%, it might be 50%, it might be 30, depending on your requirements. And if you have a mission critical environment that is active, active and redundant and all these things going on, then maybe your utilization won't be very high, but that's as high as you can make it and meet all your obligations. If a dev test environment, you can run it a lot higher. So there is no one right answer for what the best utilization is. It's kind of, it depends on your workloads and what the environment's supposed to be doing. But universally in the cloud we find it's just terrible because they rush things into the cloud without having all the maturity around it to figure out how to optimize it. Andrew, does that mean then the common mistake is underutilization? Do people are just running a lot of instances without actually knowing what's running in them or how much it's costing them? There's underutilized, there's deadwood for starters. And that's kind of a different problem. It's not that they don't know what they're doing, it's somebody forgot them. So there's no process around that. There's no ITSM, you know, process to turn these things off. So we find a lot of that. Then there's the stuff that's not utilized very well at all that you could just be running better because somebody said I need an extra large and never revisit it. The other fascinating one that we do and we find this quite a lot lately is what we call modernization. So you look okay, but you're on an old instance. There's a newer one that's a lot cheaper. You know, you're on an R3, you could be on an R4 in Amazon. So we find a ton of those and it's because people deploy an app six months ago, a year ago, and it's looked great, it still looks great, but they don't have the ability to analyze that and use benchmarks to say I have a new instance that's as powerful as that one, that's cheaper. And you need the benchmarks to do that. That's something that really doesn't happen when you have hardware, right? It's not like the server vendor calls you up and says I have a new version, I can swap out if you just tell me. Yeah, yeah, I mean, in the cloud, I give the analogy to a cell phone company. It's like they don't phone you and tell you they have a new plan that would be cheaper for you. You've got to kind of do that on your own. And so we do that for our customers. It's one of the things that we do and we kind of do it for you. So we just tell you that you just make this move, this kind of lateral move to a new instance and save a ton of money. And you know, customers, they're just, they're too busy to become experts and read the news if you're wanting to figure out if new instances come up. It's just too much, right? But wait, they haven't seen these 17 announcements that Amazon had today that might have affected them. Does your tool make the change, recommend the change? How does that kind of workflow work? Yeah, it depends on the platform, how it works but we have a very high degree of automation that we enable and there's a few reasons. One is that the analysis is so precise that when it says you do this, you can just do that. So for example, on-prem, if we say move of EM, we know it's not supposed to go with those other ones for PCI compliance. We know that that won't drive up the over commit so there's a, you know, our equation has a lot of terms and means it's very precise. So when we say to do something, you can just do it and that means you can drive very high automation as a result. What kind of granularity is this happening minute by minute or hour by hour, day by day? Well, there's two levels of granularity. There's predictive and there's real time. So one of the main things we do is that we will kind of gather all the workload history and kind of learn the patterns of that and to come up with a strategic plan saying for tomorrow, do this, put the VMs in these places and then you'll leave DRS turned on. It'll do its thing but it won't do very much because we've anticipated all the workload patterns. So a lot of times we'll do the kind of the kind of daily optimization and then DRS and VRops can do their things all day long. They just don't do as much but we do also have real time. So if we see something getting hot, we will do a hot ad on it or we can do that as well. So we kind of have a combination of both predictive and reactive at the same time. Okay. How do you handle kind of your pricing of the solution? I've heard some offerings out there that it's like, oh, we're going to save you millions and we're just going to take a fraction of that. Rather that or are you a more traditional licensing? How does that all work? It's funny, the game share we found is very hard to structure. I mean, just from a, it sounds great until you try and make the contracts for it. What we do is for on-prem we do it by target and that's a physical or virtual system. That's worked really well. That's the way a lot of our customers go there. In the cloud that doesn't work because an instance could be anything from a tiny Docker container to a giant X1. So it's as a percentage of spend. That's kind of what a lot of vendors kind of settle on in the cloud world. But we kind of don't make it infinitely variable. We know people want, they want the kind of predictability so we kind of say you're in this band it's just going to cost this and you can do whatever you want. And you know what, do you have any kind of standard rule of thumb as to you know if you have X kind of spend in the cloud we'll usually save you X percentage or and wait, if you save them a lot more doesn't that mean you're pushing them down into a lower tier? So you know, how does that get sorted out? It's a great question, that's the hazard of it but it doesn't hold us back from wanting to optimize. So what we find is if you just take for example, right sizing in the cloud if you say you're underutilized we can make you smaller in the same class a lot of people would say 15% you might save. We find that the ability to go between instance classes so again, you're memory optimized you're making you be compute optimized or vice versa, we found it's 35 to 40% pretty reliably in customers, in our customers so it's a pretty, more than page for the product many times over it's pretty compelling and it's pretty easy to get to and it affects next month's bill which is the biggest thing, you know on-prem is sunk cost, you can optimize it a lot but it's until the next refresh you may not realize the gains but in the cloud, next month's bill will actually be smaller so we find it's a lot of urgency to do it in the cloud. Kierce, what are you seeing from customers these days between their on-premises environment and the public cloud? One thing that struck me for years is if I bought gear and I'm not getting the results of the utilization out of it that kind of got a lot of attention. When I go see the public cloud there's plenty of customers like oh what do you know, I was spending three X more than I expected, ha ha I guess I'll fix it later and I was like wait if you were buying hardware you would have fired somebody and beat up your sales rep and things like that but public cloud seems to be less mature in that standpoint are you seeing that changing or what are you seeing from customers? I think there is a realization that kind of sticker shock is people where it is kind of three times more than they thought it would be but to your point there's also not really anybody whose problem that is a lot of time so we do see that becoming someone's problem like cloud architects where we see kind of more roles that are financial optimization in the cloud so people do care so I think that's a pretty positive thing I think when a lot of DevOps groups start using Amazon for the first time it's a bit of a wild west and they get agility but nobody's really looking over their shoulder I think that's starting to change pretty quickly. Yeah I wonder one of the problems I've heard I've talked to plenty of customers they're like I have to dedicate an engineer to pricing when it comes to the cloud do you solve that? Do they still need to kind of have like a dedicated person or part of a person or is that part of the value that you offer? Well that's a good question it depends on the customer size I think so we see really small organizations and again the beauty of our new offering is that you can go to really small companies or really huge companies we have customers with 100,000 systems and some with 500 and the smaller ones they may not have a big team so they may not have those roles so some of our smallest ones we're just that role for them they don't have a person that's dedicated to that kind of thing they just wait for our advisors to chime in we actually have a human advisor that's part of our service that gives you advice or insight into what's happening so for the small ones that can be that person for larger companies like the big banks that are customers of ours we kind of become one of the team so you probably still have people with a lot of these expertise but maybe you don't need to rely on it so much maybe you can not have it at all but it's more like we're someone that makes their life easier so they can go on and focus on what they should be doing which is not looking at the cloud pricing every morning Nice I see that more and more right that you need it's a service you're delivering right and that's not just bits and bytes that's customer success and you have people there that can help this stuff is crazy complicated especially if you're say a VMware admin just getting into Amazon the pricing like we just said right the pricing is very complicated so can you talk a little bit about from the admin standpoint the we realize integration and some things like that or is this there's an admin facing piece and I suppose then there's also the cost facing piece Yeah I think there's several ways we can be used you can use us almost like middleware and the admin doesn't necessarily even need to interact we some of the cost we run as an engine that's just sit there it's getting data analyzing and then making changes but you're still using VRA for blueprints and VRO and that kind of comes through us but it's kind of behind the scenes so it's a nice use case because it just adds value without making anybody's life more difficult we do have consoles that are very powerful so if you're a capacity manager or a data scientist or a cloud architect you can start logging in and actually seeing workload curves and stuff so we have some use cases where we are our interface is used quite heavily and somewhere it kind of sits in the behind the scenes and so for administrators again it tends to make your life better without making it worse again they're really busy as well and they don't as they have time to look at that So if you have a big investment to be realized that's great it just sits behind the scenes tools you already know Yeah exactly we just pull data from it and we push our else back we pull rules from DRS we push new rules down the DRS it's all very clean and so it just makes it all better without overlapping and again it makes the environment calmer so what we see in a lot of environments is you'll be able to fit a lot more work into it and you don't have the emotion activity during business hours so we're starting to measure that in our customers because volatility is an important thing like if you motion at noon at the peak of your app being busy it's not good right so we actually cause that to go away And how much of your business is on on premises and virtualized environment and versus cloud and any kind of you know line up as to where you spend the most time in the cloud Well I think for, I mean we have a lot of customers that are mostly VMware I think a good portion of them are looking at cloud in some way some of our newer customers are 100% in the cloud so that's kind of more because this is a newer offering and that's why it's quite new I think that's a smaller number right now but as far as what we're chasing down it's big it's a very large portion of it so I think it's really where we see where things are going again it's we usually do both but the cloud stuff has really captured their imagination that's what they want to be doing Any commentary on the VMware on AWS the stuff that we've heard so far? Well I mean I think it's cool it's great it's another option what I like about it is that what we find is when we analyze there's technologies that over commit and ones that don't so I can take a workload and put it in a VMware environment and over commit it and let the patterns match up and get efficiency if I put that in Amazon in like a by a large instance I might be wasting my money because I'm not using the whole instance so and I can't run a hypervisor in one of those but what we found is that for certain transactional applications it's much better to stack them together for like batch workloads it's better to run them in a rent a large for an hour so I think it's a great offering because for certain workloads it is quite efficient for other workloads it's not and we have we were showing here today the ability to analyze and compare the two saying if you took that app and put it in the new VMware on Amazon versus standard small meme and large what's the cost difference? It's a cool analysis because it's different for each app right? If I saw right their free trial available on your site is there anybody that ever comes up tries your stuff and doesn't have something that saves them money? No, we have a very good success when they try it because it's partly because it's so easy I mean it's just it's 15 minutes you pull down a connector for VMware and you plug it into vCenter or vRops and the data goes up and then we just do it all for you and it'll always find something you didn't know or some savings or some hidden risks in there usually a lot of savings hardware savings or software savings we will optimize the software licensing and in the cloud it uncovers all kinds of stuff we see all kinds of crazy stuff utilization very low so it's a yeah I've run across people that do this similar sorts of things at least at a high level on the virtualized side and on the cloud side I'm not sure that I've seen anybody that does it at both is that one of your differentiations? How do you how do you line up? What's the competitive landscape look like? Yeah, doing both is a big part I think on each of them individually we also do it much deeper so like I said in the virtual environments our ability to play Tetris with the workloads is nothing else really really like it it's we put a lot of R&D on that and in the cloud there's a lot of focus on the cost but not necessarily digging deeper into what's the cause of that cost or your Kubernetes environment you know the utilization of those nodes it so it requires deeper analytics than a lot of vendors actually have so yeah Do you give any advice as to them saying I'm trying to decide if I want to do it on premises or in the cloud do you give any guidance that way? I don't think there's any standard answer we don't we try to take sides like the data talks yeah and it's not my opinion is not an area for opinion it's just the numbers will tell you what's best for your apps and everybody's different you're talking so you know I've got this batch application oh well heck I can run this and you know some extra large thing in the cloud and you know therefore it would cost me this versus you know standing up some server farm yeah what we find is that the the the only real trick is that like absolutely if you have something that's by for 12 hours and then off for a week granting an instance for 12 hours is the way to go but the other consideration it goes back to one of your earlier questions is is multi cloud and how many different providers you want because we'll analyze the environment and that app might be cheaper in Azure and that one might be cheaper in Google and but you're not going to put each app and so you're going to choose one or two and kind of send them all there so the analytics understand that as well they're saying well you're not going to spread stuff everywhere we're going to find the best overall answer for your portfolio of workloads that that's that's an important thing okay so last question for you the virtualization admins out there is there anything that they're still doing kind of very wrong that you know that that would make their environment more efficient well I think I mean it's funny that we still see an awful lot of spreadsheets out there I it's it's it's it's funny when people try and do the numbers like to figure out where to put a new app and they'll they'll still kind of figure that out in a very rudimentary way when again science will tell you that so you can make that happen automatically so there's still certain things that people are doing manually that don't need to be done manually anymore and maybe it's their comfort zone you know maybe it's admins maybe it's other groups but I think you know again what our focus is saying that's that's great let's let's take your policies and your rules we'll just embed them and code them codify them and then you can move on to better things then then updating a spreadsheet or generating reports to send to your team every week you know like it's we have very powerful reporting so you can just make that happen and automatically to people and so it's getting out of those kind of tasks that people have done for years and moving up the value chain and saying now I'm going to focus on on cloud or on vSAN or whatever it is people want to be doing next all right Andrew Haley I appreciate you giving us all the updates on your company and I look forward to hearing more in the future John Troyer and I will be back with lots more coverage here from VMworld 2017 you're watching theCUBE