(upbeat electronic music) >> Announcer: Live from Vancouver, Canada, it's The Cube, covering OpenStack Summit North America 2018. Brought to you by Red Hat, the OpenStack foundation, and its ecosystem partners. >> Welcome back to The Cube's coverage of OpenStack Summit 2018 here in beautiful Vancouver, British Columbia. I'm Stu Miniman, with my co-host John Troyer. Happy to welcome back to the program, it's been a couple of years, actually, Boris Renski, who is the co-founder and CMO of Merantis And also is on the keynote stage for the OpenDev part of this show here. Boris, great to see you, thanks for joining us. >> Good to see you guys, and great to be back. Thank you for having me back. >> Absolutely, so we're going to talk about OpenDev, we're going to talk about a few things, but let's start with Merantis, your company. I think back to some of my first experiences at the OpenStack show. First of all, Merantis always does great keynotes, I remember there was dancing on stage, there's fun T-shirts I actually coveted. I don't go after swag much, but it was like the Heisenburg 99.999%-- >> I remember that T-shirt, yeah. >> Pure T-shirt for the Breaking Bad fans out there, to date myself on this, but always bring some energy and excitement and Merantis was one of the companies really super glued to OpenStack, so bring us up to 2018. When I think of Merantis, what should I be thinking of and let's get into it from there. >> Yeah, so let me see. We are still super glued to OpenStack. We did go through some changes and some evolutions. I think given how long it's been since we've talked, the notable changes have been a change to our delivery approach and with it some of the changes to actually the underlying software stack, so the most common thing is that we've evolved Merantis OpenStack into what we now call Merantis Cloud Platform and the key difference is how we approach actually the life cycle management of the OpenStack itself. Before our tool for installing and basically updating OpenStack was Fuel which was very prescriptive and monolithic type of delivery method and what we realized is most of it, large customers that we have, they have a fairly heterogeneous reference architectures that you have to cater to and you have to be able to do that in such a way that it is cost effective, so we've rebuilt Fuel for to a new tool called DriveTrain which uses a continuous delivery pattern to manage and deliver updates to OpenStack and with that we've also tweaked out delivery model a little bit. Before we just followed traditional distro-model where we just throw out our software out there. You can download and play with it and call us and we'll support you. When it comes to complicated distributive systems like OpenStack, that are life-cycled following a continuous delivery pattern, most of the companies simply don't have the in-house talent and skills to just take it and start deriving value, so we've moved to what we refer to as a build, operate, transfer model where we actually come in and we set up the environment, we manage an environment to an SLA, give a customer four nines SLA on the up time of the OpenStack environment we're managing and after a period of a year, give the customer an opportunity to gradually take over the operations and by operations I mean, patches, updates, et cetera until after some time we just completely go away or we just take a role of a software support vendor, effectively. So that's on the core business side. Since we haven't talked in a while, so it's a little bit of a long update, sorry. >> Stu: Yeah, yeah, it's okay. >> The thing that we've been talking a lot about recently has been the new thing we launched in beta about a month and a half ago called Merantis application platform, so Merantis Cloud Platform is OpenStack, is our core business. Merantis Application Platform is a new thing that we have launched about month and a half ago that is based on Spinnaker and Spinnaker is this continuous delivery open source tool that's been built by Netflix, originally. >> Yeah, so before we get into the OpenDev and Spinnaker and all that stuff, want your viewpoint on the OpenStack piece, so really appreciate that update. There were years that we thought, oh, it's the battle for who's going to do distributions and as you said, it's not that easy and maybe we had poor expectations as an industry as to where we could take it and where it should be used, so how should people be thinking about OpenStack in general? Can you give us one or two of the key use cases you see in your customer base? >> Yeah, so, I think that what we realized is that when it comes to general purpose cloud, so to speak, there is not tremendous value, at least among the customers that we have the opportunity to interface with, to use OpenStack. You have something that's already in place and you don't touch it and that's usually VMware or you want something new general purpose, people go to public cloud, but there is an enormous opportunity for what we refer to as tuned stacks or clouds that are tuned to particular business use cases and this is where I think is an opportunity for OpenStack to excel and this is historically where we as Merantis been actually delivering value to our customers. So speaking of the use cases, our customer base is split, we split it into enterprise and telco. More than half of the customers, actually, are from the telco side. So telco clouds, there is a variety of use cases. Typically those use cases are function of the, and the overarching use case is NFE, virtually network function virtualization. The specificity and the reference architecture of the actual infrastructure environment is a function of the VNF that is running on that cloud and in some instances if you were to categorize this for telco space, you can think of it in terms of a big cloud for VNFs that don't need to be close to the edge and those that are stretching out to the smaller footprint all the way to the edge and those are vastly different reference architectures and you do different performance optimizations and tuning and this is something that you can only do with something like OpenStack. Now when it comes to the enterprise side, the actually emerging use case that we've been seeing quite a bit of is HPC, because, again, HPC is full of purpose-built equipment, you do networking differently, you do a lot of things differently and a lot of the times the general purpose public clouds don't work for it, so for HPC again, we have a set of reference architectures that are modeled within the Drivetrain that we can just deploy fairly easily out of the box that cater specifically to the HPC use case and the enterprise. >> Boris, do you think HPC then either includes now or evolves into ML and AI as well, again, bespoke hardware, very specific use case? >> Yes, eventually. I think that there is an opportunity there for some of the reference architectures and deployment topologies currently used for HPC to evolve towards some of the AI use cases. Again, I think that, when it comes to enterprise and AI, it's a bit early, so yeah. >> Boris, the tagline of the company is, The Managed Open Cloud Company, and you talked about managing, being a managed cloud. That's been a fascinating development over the last few years. We're seeing it at the OpenStack level and for instance at the kubernetes level as well. Can you talk a little bit about that approach and who are the customers that need that entry ramp or accelerator for these private cloud installations? >> Yeah, yeah, yeah. I think that... There are two types of ways to implement infrastructure, implement the cloud. There is those that are trying to, they are looking at public cloud and they are saying, okay, this is like, I see what Amazon's doing, what Google's doing is great. I want the same thing and I want it in-house, for security reasons, for whatever, compliance reasons, doesn't matter. So all of these guys that fall into this category, I think for them to become successful with the cloud on-prem, should follow the managed approach. Again, I'm a little bit biased on this in that I'm selling this-- >> That was always the hit against running your own private cloud is you didn't have, one did not have the expertise in-house-- >> Boris: Yeah, that's exactly correct. >> That's what we need. >> First of all, the whole evolution between Fuel to Drivetrain and using the CD pattern to life-cycling the infrastructure stack is something that there isn't talent out there, there isn't DNA out there and enterprises simply are not able to just go ahead and start doing it and the whole model that, when you go to Amazon, you just have this cloud that is continuously updated for you, you don't have to worry about anything, so this model implies that you focus on delivering the end service rather than delivering the software. When you go to Amazon, you don't get software, you don't get to pick and choose. You just get certain reference architecture that is delivered for you. The guys that want to replicate the Amazon on-premise effectively, in my view, have to be gradually on ramped onto that. You can't just grab the software, do DIY, and expect you'll have an Amazon. There's a second category and the second category is basically like the software guys, the guys that, they are not looking for Amazon, they are looking for cheaper VMware, which is a different experience. I have my own team, I have my opps guys, VMware is great, but it's too expensive, I don't want be locked into it, give me something that is different. So there is value in that, but this is not the segment of the market that we are going after and I don't think that cheaper VMware is what most people refer to when they talk about cloud. So I hope that answers the question. >> Absolutely, so you brought up Spinnaker before. Want to get your thoughts on the things usually, typically on top of OpenStack, but kubernetes, Spinnaker, containers in general. What's Merantis' position on this. What are you hearing from your customers and would love to tease out some of the Spinnaker stuff a bit more. >> Yeah, yeah. Spinnaker thing is fairly new for us. We've been tracking the space and Spinnaker in particular, probably for a year, although have come out publicly just recently about it. The reason why the space was interesting to us was because I think that everybody who is undergoing digital transformation and embracing cloud as a byproduct of it, is really after being able to run the company like a startup, being able to release faster, being able to release more often and in fact, when we'd come to our customers our opening pitch even for OpenStack has always been, buy OpenStack, that'll help you build software faster. On the one hand, it's kind of like a cool pitch, on the other hand, I think everybody in the company, including myself, we're not entirely comfortable with making that leap. OpenStack means I can have an API for my VM's and maybe containers, release software faster. How do you connect the two, right? So, we decided to, in trying to solve this problem of helping companies release software faster, for once rid ourselves of our existing business and our infrastructure centric views of the world and unpack the problem and see what are the real big issues with releasing software faster today. What we realized is that one of the biggest bottlenecks is actually the continuous delivery part because when it comes to continuous delivery or even not to use fancy terms just to, deploying anything to production in the enterprise. It's a very complicated process that requires coordination between multiple teams like the application team, the SRE team, the SEC opps team, all of these teams are using different tools and the handoff process and the handshakes between are very loose, generally so a developer can build something very quickly, but for it to hit production environment, and for the enterprise to actually get feedback from the customers on this, it takes a very long time. So we started thinking about how do you actually shorten that cycle? What can you do? With that kind of frame of mind, we've come across Spinnaker and what we realized is that Spinnaker is actually, in a sense, to continuous delivery what OpenStack is to infrastructure, because the reason why OpenStack became popular is because it's effectively, on one hand, has all these plugins for diverse infrastructure, and on the other hand you can automate the orchestration process of bringing up a VM, instead of having your server people come in, put in the server, your operating people come in and install operating system, the network people come in, configure the network, et cetera, it's actually built a workflow and orchestrated the whole thing automatically without necessarily requiring companies to throw away their existing infrastructure investment. And if you go to the CD space, the situation's kind of similar. You have all these different teams, you have all these different tools, and you need to find a way to automate and orchestrate this process so that you minimize the number of human steps and this is exactly the problem space that Spinnaker's been tackling, so it's a portent of this plugability and having a single API for the entire CD chain and the best implementation would be the one like Netflix has is where the actual developers are able to just deploy to production directly. All of this orchestration between all the testing and all the stuff is done by Spinnaker behind the scenes, so we feel that actually tackling that problem and bringing this innovation into the enterprise is going to be something very dramatic at producing something at an order of magnitude performance gains for our customers. >> Of course, one of the things the foundation announced was the Zule CI/CD. Can you help us reconcile Zule and Spinnaker? >> Zule is from what I would characterize it, primarily deals with VCI side of the spectrum and I mentioned this in my talk, so one of the things we learned as a company is if you unpack CI/CD, which most people, at least in the infrastructure space look at it like it's one thing, like oh CI/CD thing, it's like one thing, basically. In reality, it's not one thing, it's completely separate things, so CI primarily has to do with actually building the code into something that can be deployed, into some deployable artifact and CD takes on from there. So Zule deals primarily with the CI part and it deals with it in a particular way for a set of specific use cases, so Zule emerged as the CI infrastructure for OpenStack Project itself and OpenStack is a very peculiar project in that, there's thousands of developers with different viewpoints on the world that are highly distributed, building many different components that are loosely coupled that all need to come together somehow. So you need to have distributed CI systems that talk to each other and you can merge all of this code and test it all together, so that use case is very relevant for large open source projects and it's probably relevant for enterprises who want to adopt similar type of practices for software development internally, so if you want to some extent de-silo many distributed Dev teams that you have internally as an enterprise and overlay standard process for the CI piece of it for everybody, I think Zule is a good solution and Spinnaker then comes after that, as an additive that does the deployment part. >> John: Yep, that makes sense. >> Alright, for us unfortunately we're running low on time, not going to have much time to dig in to the OpenDev piece. Last question I actually wanted to ask you is what do you say to the naysayers out there. People that aren't here sometimes tend to throw stones at OpenStack failed, OpenStack is dead, all the VCs pulled out years ago. Merantis has been through it and you've got customers. We've had a good experience this week, but it's a different OpenStack than it was a few years ago, so just if you could give us the final word on that. >> Yeah, so, good question. I think that... Basically, OpenStack was at this insane hype back in the day and it's natural to expect that the higher the hype, the bigger going to be the drop, but I think that all technologies ultimately, they can not sustain the hype. You have to level out at a certain point that is equal to the true customer value that you are delivering. So I think that the naysaying is a function of very high hype that has now leveled to the... What it should be, really, in terms of the value being delivered by OpenStack. And there's this pool, it generated this big pool of the naysayers that are walking around and saying that it is dead and the reason why there's the pool is because indeed there is a lot of investment, there is enormous amount of startups that kind of like, ah, we are the cool guys, we are going to change the world, we are going to kill Amazon, whatever, that now are completely gone and now of course they are naysayers and saying that the whole thing's dead, but on the flip side of that, if you just walk around the summit, you can see that there's many more users, there's many more customers that are actually talking about real use cases and then the companies that did stay and stick around, like ourselves, like Red Hat, like Canonical and SUSE, actually, are seeing continued growth and increased usage, so just a nice closing comment is our biggest customer for OpenStack is AT&T. We've been with them for five years now and they've been very excited about it and then, no it's all going to be dead, it's going to be containers now, and nuh nuh nuh, but despite all of that, the usage is continuing to grow and there is 10,000 nodes plus now running physical servers with OpenStack and it continues to work and it just, workloads are moving to it and AT&T is not the only one. There is plenty more that are following this trend, so it's a very long answer to your question, but I remain optimistic. For us it's still very much core of our business and we're continuing to see growth and usage and we are sticking around and sticking to OpenStack. >> Alright, well Boris Renski, it's, as you know, one of our earliest taglines was helping to extract the signal from the noise. We appreciate you helping us to understand the reality outside the hype. So for John Troyer, I'm Stu Miniman, more coverage here from the OpenStack Summit 2018 in Vancouver. Thank you for watching The Cube. (upbeat electronic music) (soft piano music)