>> Live from New York, it's theCUBE, covering theCUBE New York City 2018. Brought to you by SiliconANGLE Media and its ecosystem partners. >> Hello, everyone, welcome back. This is theCUBE live in New York City for theCUBE NYC, #CUBENYC. This is our ninth year covering the big data ecosystem, which has now merged into cloud. All things coming together. It's really about AI, it's about developers, it's about operations, it's about data scientists. I'm John Furrier, my co-host Dave Vellante. Our next guest is Brent Compton, Technical Marketing Director for Storage Business at Red Hat. As you know, we cover Red Hat Summit and great to have the conversation. Open source, DevOps is the theme here. Brent, thanks for joining us, thanks for coming on. >> My pleasure, thank you. >> We've been talking about the role of AI and AI needs data and data needs storage, which is what you do, but if you look at what's going on in the marketplace, kind of an architectural shift. It's harder to find a cloud architect than it is to find diamonds these days. You can't find a good cloud architect. Cloud is driving a lot of the action. Data is a big part of that. What's Red Hat doing in this area and what's emerging for you guys in this data landscape? >> Really, the days of specialists are over. You mentioned it's more difficult to find a cloud architect than find diamonds. What we see is the infrastructure, it's become less about compute as storage and networking. It's the architect that can bring the confluence of those specialties together. One of the things that we see is people bringing their analytics workloads onto the common platforms where they've been running the rest of their enterprise applications. For instance, if they're running a lot of their enterprise applications on AWS, of course, they want to run their analytics workloads in AWS and that's EMRs long since in the history books. Likewise, if they're running a lot of their enterprise applications on OpenStack, it's natural that they want to run a lot of their analytics workloads on the same type of dynamically provisioned infrastructure. Emerging, of course, we just announced on Monday this week with Hortonworks and IBM, if they're running a lot of their enterprise applications on a Kubernetes substrate like OpenShift, they want to run their analytics workloads on that same kind of agile infrastructure. >> Talk about the private cloud impact and hybrid cloud because obviously we just talked to the CEO of Hortonworks. Normally it's about early days, about Hadoop, data legs and then data planes. They had a good vision. They're years into it, but I like what Hortonworks is doing. But he said Kubernetes, on a data show Kubernetes. Kubernetes is a multi-cloud, hybrid cloud concept, containers. This is really enabling a lot of value and you guys have OpenShift which became very successful over the past few years, the growth has been phenomenal. So congratulations, but it's pointing to a bigger trend and that is that the infrastructure software, the platform as a service is becoming the middleware, the glue, if you will, and Kubernetes and containers are facilitating a new architecture for developers and operators. How important is that with you guys, and what's the impact of the customer when they think, okay I'm going to have an agile DevOps environment, workload portability, but do I have to build that out? You mentioned people don't have to necessarily do that anymore. The trend has become on-premise. What's the impact of the customer as they hear Kubernetese and containers and the data conversation? >> You mentioned agile DevOps environment, workload portability so one of the things that customers come to us for is having that same thing, but infrastructure agnostic. They say, I don't want to be locked in. Love AWS, love Azure, but I don't want to be locked into those platforms. I want to have an abstraction layer for my Kubernetese layer that sits on top of those infrastructure platforms. As I bring my workloads, one-by-one, custom DevOps from a lift and shift of legacy apps onto that substrate, I want to have it be independent, private cloud or public cloud and, time permitting, we'll go into more details about what we've seen happening in the private cloud with analytics as well, which is effectively what brought us here today. The pattern that we've discovered with a lot of our large customers who are saying, hey, we're running OpenStack, they're large institutions that for lots of reasons they store a lot of their data on-premises saying, we want to use the utility compute model that OpenStack gives us as well as the shared data context that Ceph gives us. We want to use that same thing for our analytics workload. So effectively some of our large customers taught us this program. >> So they're building infrastructure for analytics essentially. >> That's what it is. >> One of the challenges with that is the data is everywhere. It's all in silos, it's locked in some server somewhere. First of all, am I overstating that problem and how are you seeing customers deal with that? What are some of the challenges that they're having and how are you guys helping? >> Perfect lead in, in fact, one of our large government customers, they recently sent us an unsolicited email after they deployed the first 10 petabytes in a deca petabyte solution. It's OpenStack based as well as Ceph based. Three taglines in their email. The first was releasing the lock on data. The second was releasing the lock on compute. And the third was releasing the lock on innovation. Now, that sounds a bit buzzword-y, but when it comes from a customer to you. >> That came from a customer? Sounds like a marketing department wrote that. >> In the details, as you know, traditional HDFS clusters, traditional Hadoop clusters, sparklers or whatever, HDFS is not shared between clusters. One of our large customers has 50 plus analytics clusters. Their data platforms team employ a maze of scripts to copy data from one cluster to the other. And if you are a scientist or an engineer, you'd say, I'm trying to obtain these types of answers, but I need access to data sets A, B, C, and D, but data sets A and B are only on this cluster. I've got to go contact the data platforms team and have them copy it over and ensure that it's up-to-date and in sync so it's messy. >> It's a nightmare. >> Messy. So that's why the one customer said releasing the lock on data because now it's in a shared. Similar paradigm as AWS with EMR. The data's in a shared context, an S3. You spin up your analytics workloads on AC2. Same paradigm discussion as with OpenStack. Your spinning up your analytics workloads via OpenStack virtualization and their sourcing is shared data context inside of Ceph, S3 compatible Ceph so same architecture. I love his last bit, the one that sounds the most buzzword-y which was releasing lock on innovation. And this individual, English was not this person's first language so love the word. He said, our developers no longer fear experimentation because it's so easy. In minutes they can spin up an analytics cluster with a shared data context, they get the wrong mix of things they shut it down and spin it up again. >> In previous example you used HDFS clusters. There's so many trip wires, right. You can break something. >> It's fragile. >> It's like scripts. You don't want to tinker with that. Developers don't want to get their hand slapped. >> The other thing is also the recognition that innovation comes from data. That's what my takeaway is. The customer saying, okay, now we can innovate because we have access to the data, we can apply intelligence to that data whether it's machine intelligence or analytics, et cetera. >> This the trend in infrastructure. You mentioned the shared context. What other observations and learnings have you guys come to as Red Hat starts to get more customer interactions around analytical infrastructure. Is it an IT problem? You mentioned abstracting the way different infrastructures, and that means multi-cloud's probably setup for you guys in a big way. But what does that mean for a customer? If you had to explain infrastructure analytics, what needs to get done, what does the customer need to do? How do you describe that? >> I love the term that industry uses of multi-tenant workload isolation with shared data context. That's such a concise term to describe what we talk to our customers about. And most of them, that's what they're looking for. They've got their data scientist teams that don't want their workloads mixed in with the long running batch workloads. They say, listen, I'm on deadline here. I've got an hour to get these answers. They're working with Impala. They're working with Presto. They iterate, they don't know exactly the pattern they're looking for. So having to take a long time because their jobs are mixed in with these long MapReduce jobs. They need to be able to spin up infrastructure, workload isolation meaning they have their own space, shared context, they don't want to be placing calls over to the platform team saying, I need data sets C, D, and E. Could you please send them over? I'm on deadline here. That phrase, I think, captures so nicely what customers are really looking to do with their analytics infrastructure. Analytics tools, they'll still do their thing, but the infrastructure underneath analytics delivering this new type of agility is giving that multi-tenant workload isolation with shared data context. >> You know what's funny is we were talking at the kickoff. We were looking back nine years. We've been at this event for nine years now. We made prediction there will be no Red Hat of big data. John, years ago said, unless it's Red Hat. You guys got dragged into this by your customers really is how it came about. >> Customers and partners, of course with your recent guest from Hortonworks, the announcement that Red Hat, Hortonworks, and IBM had on Monday of this week. Dialing up even further taking the agility, okay, OpenStack is great for agility, private cloud, utility based computing and storage with OpenStack and Ceph, great. OpenShift dials up that agility another notch. Of course, we heard from the CEO of Hortonworks how much they love the agility that a Kubernetes based substrate provides their analytics customers. >> That's essentially how you're creating that sort of same-same experience between on-prem and multi-cloud, is that right? >> Yeah, OpenShift is deployed pervasively on AWS, on-premises, on Azure, on GCE. >> It's a multi-cloud world, we see that for sure. Again, the validation was at VMworld. AWS CEO, Andy Jassy announced RDS which is their product on VMware on-premises which they've never done. Amazon's never done any product on-premises. We were speculating it would be a hardware device. We missed that one, but it's a software. But this is the validation, seamless cloud operations on-premise in the cloud really is what people want. They want one standard operating model and they want to abstract away the infrastructure, as you were saying, as the big trend. The question that we have is, okay, go to the next level. From a developer standpoint, what is this modern developer using for tools in the infrastructure? How can they get that agility and spinning up isolated, multi-tenant infrastructure concept all the time? This is the demand we're seeing, that's an evolution. Question for Red Hat is, how does that change your partnership strategy because you mentioned Rob Bearden. They've been hardcore enterprise and you guys are hardcore enterprise. You kind of know the little things that customers want that might not be obvious to people: compliance, certification, a decade of support. How is Red Hat's partnership model changing with this changing landscape, if you will? You mentioned IBM and Hortonworks release this week, but what in general, how does the partnership strategy look for you? >> The more it changes, the more it looks the same. When you go back 20 years ago, what Red Hat has always stood for is any application on any infrastructure. But back in the day it was we had n-thousand of applications that were certified on Red Hat Linux and we ran on anybody's server. >> Box. >> Running on a box, exactly. It's a similar play, just in 2018 in the world of hybrid, multi-cloud architectures. >> Well, you guys have done some serious heavy lifting. Don't hate me for saying this, but you're kind of like the mules of the industry. You do a lot of stuff that nobody either wants to do or knows how to do and it's really paid off. You just look at the ascendancy of the company, it's been amazing. >> Well, multi-cloud is hard. Look at what it takes to do multi-cloud in DevOps. It's not easy and a lot of pretenders will fall out of the way, you guys have done well. What's next for you guys? What's on the horizon? What's happening for you guys this next couple months for Red Hat and technology? Any new announcements coming? What's the vision, what's happening? >> One of the announcements that you saw last week, was Red Hat, Cloudera, and Eurotech as analytics in the data center is great. Increasingly, the world's businesses run on data-driven decisions. That's great, but analytics at the edge for more realtime industrial automation, et cetera. Per the announcements we did with Cloudera and Eurotech about the use of, we haven't even talked about Red Hat's middleware platforms, such as AMQ Streams now based on Kafka, a Kafka distribution, Fuze, an integration master effectively bringing Red Hat technology to the edge of analytics so that you have the ability to do some processing in realtime before back calling all the way back to the data center. That's an area that you'll also see is pushing some analytics to the edge through our partnerships such as announced with Cloudera and Eurotech. >> You guys got the Red Hat Summit coming up next year. theCUBE will be there, as usual. It's great to cover Red Hat. Thanks for coming on theCUBE, Brent. Appreciate it, thanks for spending the time. We're here in New York City live. I'm John Furrier, Dave Vallante, stay with us. All day coverage today and tomorrow in New York City. We'll be right back. (upbeat music)