 This is theCUBE, our flagship program. We go out to the events and extract the signal from the noise. I'm John Furrier, the founder of SiliconANG. I'm John Mike Ho. Stu Miniman from wikibond.org, and we're here with Ranga Charri, who's the vice president general manager of storage, big data for Red Hat. Welcome to theCUBE. Thanks, John. Thanks. That's a lot of action going on, storage, big data. I mean, that's where all the action's at. And we love covering storage. We love looking at it. We love covering storage. We look at storage. And storage is always the conversation it always comes down to, because at the end of the day, you got to store the data. And it's got to fit into the operating system. So it is a subsystem of the operating system. But now in context to the cloud, it's a huge conversation. And I want to get your thoughts on the kinds of things that you're seeing at the show, what you guys are talking about, and how does the storage fit into the software-defined data center, into the cloud, on the public, private, with virtualization, and the physical world? Because that's a key component. It's a huge subsystem in everyone. It's re-architecting their operating environments. Absolutely. There's a lot of things that we can cover. But first off, thank you. Thank you for having me here. My only suggestion would have been have this outdoors. It's a beautiful day out there. It'd have been great to bask in the 60 degrees temperature, but it's beautiful. But you had a lot of questions. But the first and foremost, I think, what I've been, this is my fourth summit. And what I'm really impressed and what Jets gets me excited is the continued innovation. You know, when I talk to customers, talk to partners, every year you go back and there's a new thing in the horizon. There's new leading indicators of how people are trying to take advantage of some of the technology advances to solve their business problems. Getting into storage, you know, I think what we saw was a pronounced shift over the last three or four years. And I would argue that it was because of a couple of reasons, right? One is just the movement to cloud. Because if you really think about it, what cloud does, especially on public clouds, you cannot take a piece of storage hardware and move it into a public cloud, right? It defies a lot of physics. So that, in a way, really necessitated the need for a software-defined approach to storage. The other aspect is just a new type of workloads. And unstructured data. Industry analysts expect that over the next five years, 90% of enterprise data is going to be unstructured in nature. So unstructured data requires a different type of treatment. It's not no more scale-up, it's scale-out. So you've got to continue to deal with the volume and the velocity of the data that's coming in. So those are the two trends that we see are really bringing this to the forefront. So a lot of folks have been dealing with software-defined storage and kind of the older paradigm, you know, park it out there, I got a fence around, I go out and send some queries, but now as data becomes central to the value proposition, it's part of the development, but it's also part of the OS. So I want to get your definition. What is, in your view, software-defined storage? Very simply, and I think the way I define, or the way, not me, but even the customers to truly decouple the services that your software layer can provide, independent of the underlying hardware. Right, so the intelligence elevates to the software level and the ultimate goal is you should be able to take industry-standard commodity x86 servers which has got a tremendous amount of horsepower built into it, and layer in your software solution on top and just have all the services that you've been providing so this is data replication with a data backup or data migration so the entire gamut of services that you're used to, the only difference is to the customer it gives them complete flexibility and freedom of choice. I could go with an HP server today, I could go with IBM server tomorrow, I could go with a Dell server the day after and nothing changes as far as the storage services that I provide. So really able to separate out the hardware and the software is just not, you can look at it in isolation and storage, but I think the perfect chemistry is with cloud, compute is becoming software-defined. Networking, you heard about software-defined networking so the triple play here is software-defined storage, software-defined compute, and software-defined networking. So Ranga, actually Wikibon did some research earlier this year, we created a market definition of what we called a server-sand really a scale-out architecture that allows both compute and storage to be able to scale. In general, if it's done right, it should be open, it should be highly scalable and we thought Gluster actually fit under, as a service that could be part of this type of architecture, it fit into our definition. Can you tell us, Red Hat bought Gluster in 2011? It really, obviously Red Hat had a long history working with the storage ecosystem, but it did change a little bit how Red Hat looks at the storage market. Can you lay out for us Red Hat's vision of the storage portfolio? Absolutely. We've had, let's say file systems since day one and I think if you go back to late 2011 when we acquired Gluster, what we saw was a pronounced shift in the customers buying patterns and customers definition of how they wanted infrastructure to be set up and the two key things was we saw just the cloud and we say cloud and using the term pretty loosely which is private cloud, public cloud, open hybrid cloud and the unstructured data. So those were the two key drivers where we said look, the old ways of doing it aren't going to cut it. So that was the starting point for us to go after this market opportunity that's out there to take advantage of. Now the other piece which we don't talk about much for the right reasons we understated is what really attracted us to the Gluster solution, obviously the technology was great but more importantly the community. I mean for a company when we acquired them was two and a half years into existence but the community was so vibrant and we felt that we could under the auspices of Red Hat under the stewardship of Red Hat we could really move the community forward and we've done that. We've done the phenomenal job I think of continuing to foster some of the innovation that's going on, not just in software defined storage and open software defined storage. So you mentioned the aspect of how Gluster fits into cloud deployments by understanding this can be kind of private or public and service providers. I haven't heard much talk about kind of the service provider and public cloud from Red Hat providing the infrastructure for them obviously customers put deployments of Red Hat everywhere. Can you speak to the service providers and the like? So the service provider market from a service provider standpoint you've got the classic service providers who provide storage as a service to their customers but you also have some of these large organizations or internal service providers. So there's some people who provide storage as a service as a basic service. The other service providers basically take storage as an integral component of the service that they provide to their customer. An example of that is Kandora Radio. Kandora Internet Radio even before the acquisition of Gluster was using Gluster as a way to archive the media and everything else. So they were actually providing storage in a way as a service to their customers. Just the other example I can think of which is a different type of storage as a service which is Intuit. I mean today as you all know in North America yesterday was a tax deadline and Intuit uses Red Hat storage as the foundational storage platform for the web properties. So when you go in and click on the Submit button that gets stored within Red Hat Storage. I want to ask about the customer attraction in context of Red Hat and where Red Hat is going. So at Red Hat obviously there's no doubt, I mean we talk to the customer all the time and they've had Red Hat Enterprise Linux for years and years powering their business all the greatness of Red Hat but they've dealt with storage. You've got EMC drives and you've got NetApp legacy infrastructure quite frankly is working pretty damn well and they're okay with it but they realize you've got to go to the cloud. So I've got to ask you how do you see the storage component that you are architecting and solutions you're rolling out fit into the vision where Red Hat is going to the cloud specifically around OpenStack, OpenShift. That's the hot area. We saw REL7 Red Hat Enterprise Linux 7 released yesterday. We've got availability with the private cloud with Dell announced today. This major attraction in OpenStack all the enterprises see OpenStack as a storage to the cloud not as a mutually exclusive environment but one that's going to coexist. So how do you look at that? You've got legacy I've got to deal with physical hardware software is the key but how are you guys building that solution? So there are three aspects of it right. One is even if I were to pop it a level higher every one of our products today whether it's storage or OpenStack or OpenShift, the central theme behind that is Open Hybrid Cloud customers on a journey of not just a hybrid cloud but an open hybrid cloud the ability for them to say look I'm going to use this public cloud today but tomorrow for whatever reason I'm going to move to a different public cloud how do you really create this multiple bridges? So having said that if you think about storage it's an integral part of it because data needs to move back and forth regardless of where it resides so storage is a very I'd say more than a critical it's an integral part of an open hybrid cloud strategy so the proof point of that is today if you look at Red Hat Storage you can run it in a physical environment to your point you can run it in a virtual environment you can run it in an OpenStack environment or you can run it in a public cloud environment and the fact is because the way Gluster which is Red Hat Storage is based on is designed to be POSIX compatible you don't have to do any application rewrites so I create an application on-prem I just move the application let's pause it for a second that's a really important point because that's a requirement that people have is POSIX compliance is a huge deal that's not like a throwaway feature so if I'm a new vendor if I don't have POSIX compliance you're pretty much specced out right? You are and it's also the fact that what it does for them is it takes a tremendous amount of the friction in creating rewriting the app to make it purpose built for cloud A, cloud B, cloud C so we've taken all the friction out of the system a good point I think to your point was that on OpenStack today's Dell announcement as well as yesterday I think we talked about a few customers publicly talking about OpenStack so one of the customers and we got a lot more but one of the customers was the University in Portugal in a great use case where they're actually using Red Hat Enterprise in OpenStack and Red Hat Storage in conjunction because in an ideal world you need both hands to clap you can't just go into here's my virtual compute what do I do for storage having them work together we might have to do a site visit to Portugal that would be a good opportunity this is interesting because what we're seeing is two threads right customer, migration to the cloud so I have an enterprise that has pre-existing infrastructure and software and the application so yesterday we heard the big theme was application delivery we heard doctrine containers and containers to cloud is what virtualization is to the OS if you want to look at kind of a parallel we talked about that yesterday you guys talk about this internally at Red Hat and within the open source community Linux kernel has always been the discussion so I'm a developer back in the old days when I was an old developer memory was a real scarce resource and disk was the viable swap out because it was abundant now with flash and persistent you have a paradigm shift going on so there's been discussion in the industry around Linux kernel, software practices where we can take advantage of not just RAM but how is that changing the storage equation, can you elaborate what you guys are talking about internally and how you guys look at that trend and how that affects the developer so from a technical standpoint a thing that it's worth repeating here is the Red Hat storage technology it runs in a user space so what that means is there's no kernel implications so we take advantage of any new kernel functionality that's out there without having to go modify the kernel to suit a purpose so architecturally having it run as a user level application frees a lot of innovation innovation really makes it possible so you're not creating a dependency exactly that's a great way to put it so the developer is looking at the kernel development they could pivot off whatever innovations happen there absolutely and there's a lot of proof points here that we can talk about but in respect to the SSD piece one of the things that we keep very close eye on is not just innovation around flash and SSDs but also if you look at it today I mean the other day I was reading that Seagate has come up with 5TB drives in production so if you think about it you can get just pick a model HP server with 60 drives concatenate 8 of them together you now have a 1.2 petabyte storage farm at I'd say 10th of the cost of what people are used to today I mean that's the beauty of parallelizing your software innovation along with innovation that's going on on the hardware side so Ranga since we've got the discussion of storage going on I'm wondering if you have commentary on what the folks over at VMWare have done with vSAN really kind of you know churned up a bunch of discussion as to storage architectures and the role of really you know virtualization so what's your take on what's happening with vSAN well so without getting into the specific the demerits of any specific solution our approach is very much different in that it's truly open so we don't force you to run one type of hypervisor to take advantage of the storage functionality if you know on our operating system side of things and people run REL on top of VMWare people run REL on top of HyperV so give the customers the choice as opposed to saying you know that's one of the problems today I see with the enterprise storage space today it's proprietary not just a proprietary it's a bunch of silos right I got a silo for application A that's SAN based I got silo B for NAS based and forget the capex the cost of operating those environments with different tools and different skill sets it's just driving these guys nuts and that's what our customers tell us they say don't force me into another silo right and if we use the tagline liberate your information and that resonates with a lot of people because now they can go and say okay the shackles have been removed I can do what I want so the data center has been a big battle ground and there's been a lot of marketing around the sub-defined data center which really drove out of as you mentioned earlier sub-defined networking and Nasir was the shot hurt around the world when they sold for a billion dollars on $50 million B.C. investment from VMware everyone kind of took you saw Cisco kind of reshuffle some things and so you know that is really the beginning of what started moving down to the emphasis of software and you mentioned that earlier so when you look at that trajectory where do you guys see the evolution of the data center in context of storage and in context of big data because you have to oversee the big data component so if I'm putting data in the cloud I have data on the data warehousing business intelligence market under under-season being disrupted with new economics visualization and data insights are the top customer driven things that they want so how do you make that low latency data modeling really become core to the software so I mean I can give you the radar perspective and also the customer's perspective you know I think when customers look at some of the leading properties like a Google or a Facebook they know they can do it in fact I had lunch with one of our customers yesterday and he said the mandate or the edict from the top is what he defined as a commodity data center right which is stop going to these specialized things I just want everything standard parts much like what the classic web facing properties are doing today and then hardware is going to fail it's got to be replaced just move on right so the intelligence is moving into the software space now with aspect to virtualization and data virtualization specifically we have a very rich product portfolio I don't know JBoss product line one of them is our JBoss data virtualization product at the core what it does it allows you to get data from sql databases from no sql databases regardless of what the myriad of sources are and have them come to one single resting point so then if that's stored on red hat storage you can run your analytics on top of that and you can take advantage of all the you know hardware advances with flash or SSDs or everything else you can get away from and even within the big data space right you're seeing some of the new emerging trends around things like spark and other things that are really starting to it's interesting you're seeing the original Linux red hat model of decoupling software from the servers now going at such a large scale and pop my worry was saying yesterday that data virtualization in her opinion is the next big thing because then you start looking at the data side of the equation where data becomes a core aspect of what's going on in the OS do you agree with that and share with the folks that aren't in the trenches who don't know all the nuances of data virtualization why is that such a big trend why is she saying that why are you feeling the same way so take the virtualization word out of it right so the way customers think about it is storage data information that's kind of the logical I need a place to put this which is my storage infrastructure they need all this data out there and if you go back to the model from 20 years ago everything was physical so I had a big huge server box with some storage in it I had a whole bunch of ETL tools and all that stuff they were pumping in data now with virtualization the added advantage of virtualization cutting across those three veins is the fact that now you can really pack a lot of horsepower into a single physical chassis or a single physical cloud and it's basically I think it's an economy of scale game and moreover economies of skill game right you don't have people going across multiple places trying to do so if you think about it and I don't know whether that's the way this is really top my head right now but storage virtualization data virtualization who knows the next step would be information virtualization yeah and also it solves a privacy issue everyone talks about Gmail Google they store all your emails it's just cheaper to store forever and just to worry about deleting it but at some point you got to blow away the data and might have some privacy issues do you want to comment on that actually I wanted to get a little more from from the big data aspect of things you know before we wrap up so you know when we think about you know red hat and big data you know other than being kind of the operating system for it you know what should we think of when we think of you know red hat so big data is more than I do I mean one of the common misconceptions is you know big data is Hadoop I think big data is more than Hadoop so if you go look at every let me give you a couple of examples right one is obviously red hat enterprise Linux today is a huge operating system where all these big data workloads are run now if you elevated to the cloud side of things for instance one of the things we did with Hadoop in this case was at an Umbari where we actually work to the community on the Umbari project to make it more versatile to support other file systems besides the HEFS we're doing something similar in the open stack with a project called Sahara which used to be called Savannah which is an orchestration of open stack so up and down the stack compute storage network application yesterday Hortonworks announced something that openshift product line the ability to do you can now scale your Hortonworks environment running on top of open stack so you'll see us participating in every part of the ecosystem of the stack up there so on the big data front what do you think about the whole validation with the cloud area Intel news because that really got everyone's attention so in the sphere of the Hadoop world that's a massive event but you know it's a small blip on the radar when you look at the entire computer industry what you guys have a purview into with red hat and Linux you're seeing Hadoop becoming a really relevant piece of that now so Intel is working closely with cloud on that investment but yet Hortonworks and we had Intel on yesterday saying hey we still love Hortonworks, we're Intel just because we make one move doesn't mean we understand ecosystems and specifically upstream Hortonworks is plugged in and Hortonworks some say we've said is the red hat of Hadoop same business model doing very very well so Hadoop is now elevated but that's not the only data open source thing so as this becomes a much more mainstream how are you guys developing that part of the ecosystem can you elaborate what's going on with Hortonworks you're doing joint engineering is it just more partnership at this point what's going on there no so specific to Hortonworks we announced two months ago I think mid-February where it's more than just Barney relationship if you will there's a lot of engineering engagements going on between the two companies but the broader point is the cool thing with big data I'm using the big word big data pretty loosely is beyond Hadoop the center of innovation around all these new big data projects are all open source that's actually a great platform for red hat, for open source and for red hat where we absolutely want to have a platform where ISVs can really thrive and regardless of whether it's Hadoop or whether it's it's a huge ecosystem on analytics right now you're seeing a lot of companies kind of finding a lot of white space in the analytics space which they're not trying to be a platform part of it so that's really a good signal from the ecosystem standpoint so I got to ask you as the platform is a service battleground continues to be a top top of mind with OpenStack and OpenShift relative to other approaches the data layer becomes a conversation now so I'd like to get your perspective how do you see the the storage data layer in OpenStack developing because that's a real key part of the stack in this middleware environment Yeah so in OpenStack or OpenShift OpenStack first and then maybe OpenShift I'd say the basic foundational stuff exists today from a technology standpoint where you want to provide file services, object services and block services but there's a lot more things in fact Red Hat and a few others we're working on a project called Manila which is file as a service for OpenStack so that's going to happen on the OpenShift side of things we are looking at to be containerized Red Hat storage things like that so there's a lot of things that are going on not just within the four walls of Red Hat but out in the upstream community Roger the general manager of Red Hat great to have you on the Cube I got to ask you one last question what are you looking at around the corner right now it's pretty clear you guys are on the straight and narrow let's get someone to find storage out you've got a big data component works pretty obvious as you slow down to take that next corner what are you watching that's around the corner that we should be mindful of as an industry and within the Red Hat ecosystem around the role of storage is it the data layer, is it the continuing Linux kernel kind of issue we talked about is there other things that we should be mindful of you know I think we look at it, I think around the corner it looks very nice when we peek it on there it looks really good because the two aspects right one is more than and I think you brought the question up earlier which is more than the enterprises the opportunity for service providers who want to start offering storage as a service to their customers and I think that part of the battle ground I think it's still not there yet so that's one of the things we are paying very close attention to we are here inside the Cube Live in San Francisco for the Red Hat Summit Ceremony our 10th year of this event the ecosystem is changing and growing evolving, as we said yesterday the mojo continues for Red Hat and again open source is now tier one it's taking a whole nother level power in this big innovation cycle we are living in the technology world this is the Cube, we'll be right back with our next guest