 Live from San Francisco, California. Extracting the signal from the noise. It's theCUBE, covering DockerCon 2015. Brought to you by SiliconANGLE Media. With special thanks to Docker. Now your hosts, Stu Miniman and Jeff Frick. Hey, welcome back everybody. Jeff Frick here with theCUBE. We're at DockerCon 2015 in downtown San Francisco at the San Francisco Marriott Hotel. About 2,000 people running around, talking containers, talking tech, it's a geeky show, a lot of fun, a lot of good energy. And they just, Stu, they just charged everybody up with a bunch of ice cream and the coffee's been nonstop. So I think it's the best practice. You know one of the measures we always have for these shows is it actually had good feud. There you go. And by the way, Jeff, we haven't mentioned they've got Legos. I mean, you know, what do geeks love? We got T-shirts and we got Legos, right? They got the Docker ship over there. Guess the number of Legos you probably win a prize. So we're excited for this next segment to have Kamal Srinivasan from EMC. He's the principal product manager for EMC Advanced Software. Welcome to theCUBE. Thank you, Jeff. Thank you, Stu. So begs the question, what is EMC excited about Docker for? What are you guys here? You got a big presence? Sure. So we are very excited to be at DockerCon this year. This is actually the second year of DockerCon and coming and seeing the energy here, not only from the attendees who are here, but the community participation coming and contributing to the different projects. EMC is very excited to be here because today we announced our elastic cloud storage to our dough, the free and frictionless download available for unlimited time, unlimited capacity, going with the other EMC trends of making our software available for wider community, right? It's not enough when we do the software and make it the best in class for that storage product, but having the wider community participation to make sure that the software can get better, that's where EMC is putting a lot of focus on. And that's what we announced today at DockerCon. Yeah, so come on, can you step back? Give us a little bit of background on you. My understanding you worked on the Azure team. Yeah. Microsoft, you're up in Seattle area, joined EMC. What brought you to EMC? You know, talk about how your journey brought you to the show. I come from sunny Seattle. It's warm and bright there, much brighter than San Francisco. And so my background is definitely in the cloud. A lot of time spent in the cloud. I was an Azure, Microsoft Azure. I was in the networking team and a few years back joined EMC and came with the group of gentlemen who started Amitabh and the group who started the advanced software division at EMC. And our strategy has been to come in here and build, make this a software-defined platform, right? And our team specifically focuses on how to make this storage software completely as we're talking of LEGOs, a LEGO-based architecture, so that the software, where the storage, the protection, the georeplication, everything, is a LEGO that's abstracted away from the southbound, the hardware, or the northbound, the access protocols. So that's what we do in this advanced software division. And my role here as a product manager, so in charge of working in the different aspects of the object taking into market, the implementation strategy, and go-to-market strategy for it. All right, so Kamal, you're wearing the Elastocloud storage shirt. Is this a new logo? Yeah, last year at EMC World, ECS 1.0 was released. Bring us up to speed where ECS is today. Sure, so we came to EMC World last year, we announced our 1.0, which was our first version of Elastocloud storage. We had shipped object platform, which was shipping as part of Viper, even prior to that. But it was running against intelligent or platform-to-erase. When we came to last year's EMC World in 2014, we said, now we've got this complete, commoditized Elastocloud storage running on commodity hardware, and we shipped it in form factor of an ECS appliance. Now, fast forward this year, one year onward, we've come with Elastocloud storage 2.0. Our strategy is to take the software as I was talking about this Lego-based architecture. We want to not only enable the EMC Commodity Appliance, but we want to enable all third-party Commodity Appliance. And going further, we want to enable wider dev test, running on your laptops, or running on any kind of commodity infrastructure where you can do scale testing without having to deploy a huge infrastructure for an object store. That's where 2.0 is. In 2.0, we've done a lot of enhancements from the product itself. We've done a lot of geo-enhancements for multi-site optimizations. So one of the things I would say is that we've been testing in the background since 1.0, with a lot of customers who've been deploying not just the internal customers, but even some of the external customers who've been deploying at scale thousands of nodes and testing multi-petabyte environments. And over here at DockerCon, we were talking about this. We were talking to a lot of customers here who have come here talking about their own Docker experience. So one thing I was learning was that I think EMC has come out with this Dockerized container for elastic cloud storage, right? And this is one of the probably the containers that's running thousands of nodes of stateful Docker in production and enterprises. So I was glad to hear that a lot of people gave positive feedback and a pat on our shoulders for having to run Docker in this fashion. The other enhancements we've done are a lot of monitoring diagnostics because we're running a cloud platform. We are enabling these enterprises and service providers to run their own cloud with the elastic cloud storage. So when they're doing that, they need insight into the system, not only in terms of performance, but how the system is keeping up, how the capacity is utilized, the network metrics, all those kind of things. So we've gone ahead and done a lot of monitoring and diagnostics, so giving them the capability of running the whole private cloud in their environment. So those are some of the enhancements that we've done. And then we've also caught up with a lot of enterprise IT readiness, a lot of service provider readiness in the system to make many of these customers go to production with ECF 2.0. All right, so Kamal, you gave a session this afternoon. How many people showed up and give us a little spotlight of what you talked about? So we ran a session on Docker tutorial. It was a tutorial session on Docker persistence. So essentially, there were two pillars on it. There were about a hundred plus people. We were, in fact, the guy who was running the session came in and said, we might be in violation of the fire code there because the room was packed and there was floor seating and standing and maybe some people on the podium, I don't know. So the session was really on how EMC is contributing to Docker in terms of persistence. So we came out with some announcements on how our EMC code team has done some cluster HQ work and how we've contributed back to the Docker community. And the second part of the session was on the elastic cloud storage and how we've run this in a Dockerized version and how it's available for customers to download and use and do DevTest. So we've made it available not just as a container, but we've put some puppet scripts around it. It's available on GitHub, on Docker Hub, so you can go download. It's emc.com, get ECS, or go to github.com, search ECS, and you should be able to find the software. Yeah, so when you look at the applications that sit inside Docker, lots of the applications are stateless and short lived, but when you start talking about stateful applications, they probably live a little bit more. Can you bring us through what you see at the application side and how does EMC make any differentiation on various type of apps? Sure. I think the whole containerization is catching up, right? As we can see around us at the DockerCon, the second DockerCon, just the energy and the number of people shown up here. So container as a service or the applications and microservices running, it's catching up. So as this is catching up and customers are trying to have these conversations on how do I move from a stateless, just taking my application running within a Docker container to making this equivalent of if you rewind back when VM started, right? And was getting into this mode of making VM the next big thing to run. There was these challenges that you had to overcome, right? Get over the helm. I think Docker is at the same place where statefulness is the first problem, one of the biggest challenges to solve. And even at the birds of feather session during launch, we were having some interesting conversations with some biogenetics company and some financial companies, kind of very different verticals, but the challenges they were having is the same thing because they have a mix of greenfield and brownfield applications. They're trying to get into that data center which is all containerized and they're trying to get it stateful. Solve the typical common challenges which are solved in a VM world. And when they come here, they're facing the problem. One of the big things is the persistence, right? Because they have to get the volume persistence for local but now when you move your container dies, your state dies. So you need to get that brought up the container and another node that container needs to have the same state. So how do you solve those problems? So with EMC, what we're contributing to this Docker community with the scale IO and cluster HQ and with ECS with this automated thing, it's giving persistence from two aspects. Persistence from a local volume or going across the scale IO and also persistence from this cloud, deploying an object store and getting your containers, being able to use this persistence of raw commodity infrastructure. Yeah, so can you talk about, you mentioned a couple of vertical use cases. Where do you see this starting out? Are we mostly really in kind of test beds at the point and maybe walk us through what would you expect to see from a maturity standpoint of persistent apps and containers? I think so the gentleman, I was having a conversation on, he was from financial vertical and he was in fact running production with one of his applications, right? So he was running like a few thousand containers in production, not just DevTest. So I think containerization is picking up not just for DevTest, but also for production, right? And it depends, the verticals, we see definitely the financial verticals picking up, some of the verticals around the research labs, the genomics, those kind of life sciences verticals, definitely taking the leap ahead into the space and getting a lot of their applications because they have more leeway to get into this containerized world with the microservices. So we see them much more moving forward ahead in comparison to some of the other verticals, but some of the other verticals are also catching up in terms of running. So for example, one of the customers we work with, they're running a completely media company and they're trying to back up to the cloud, they're running, they're building a data center which is all container-based. And at EMC World, they were on stage with CJ, this is Verizon, we're talking about how they're building this container-only data center, right? So that all their applications, including storage, could be containerized and run in this new data center, right, this gen-next data center for their applications. So it's not just green field but also ground field applications. That's great, so you're seeing it in production because we were getting some hallway chatter during one of the breaks where someone was saying, ah, you know, we do a lot of stuff besides production but we haven't really figured out the production thing, we're not really ready to cross that bridge, so we're doing a lot of other things. But it sounds like you talk to a lot of people here at the show that are in production, that are using it. Right, I would say a lot of people are getting there and some people have already gone into production, right? That's how I would categorize. And it's not without challenges. I think there's a big hump to cross. So any of the customers we are working with today, it's not without challenges that they get to where they are, even with DevTest or production. Like challenges along the networking, verticals of things. I think Docker today in the keynote, they were announcing some of the networking, how they are changing some of those things, how within the container you can get some of those similar concepts of segregation, all the traffic related to VLAN tagging, those kind of things within there, some of the aspects of how to standardize the container deployment, right? And again, there's like a lot of fabric technologies available, but which fabric technology to choose to run there? For example, we had to write our own fabric technology when we did ECS 1.0. Since I said, even with ECS 1.0, we were running in containers. And if you think about it, Docker didn't go GA at that time when we were working on ECS 1.0. So we had to pick up a pre-GA version of Docker and we placed a bet on it, right? And we said Docker is going to become big and we found some really great use cases for Docker to be the big bet for us. So when we picked up, there was no fabric, there was no misos, there was no fleet in the state where we could go and use that for our deployment and lifecycle management for containers. So we had to build our own fabric, which was doing this lifecycle management. And since then it has evolved. So now we have our own fabric, which does a lot of this lifecycle management and those kind of things. Now being AMC, we could invest in a lot of these kind of investments and engineering to get this fabric. Now that's not true for many companies because think of, this doesn't scale, right? Even if you can, investing in these kind of every single problem to get your Docker to production, I think those are the challenges companies are facing in order to get to production. Yeah, so we're still in the early days here, Kamal. One of the things I'm curious about, you guys do scalable architecture, you're leveraging Docker, can we do that over geographic distance if we're containers? Because kind of the persistence is one thing, but getting that replication and having applications span distances is usually a pretty hard challenge. I didn't think we were there yet, but I'm curious if you could clarify for us. ECS is a completely a geo-distributed, active-active, strongly consistent architecture, right? So if you think about it, the fact that we are running in containers is giving persistence from the underlying layer of the object storage, and the object storage itself is a geo-replicated, strongly consistent system. So if you think about the application running within the container, it's getting this geo-distributed nature by the fact that the underlying storage is getting this geo-distributed, strongly consistent, active-active infrastructure, right? So to quote again from the keynote, I think there was a time when it was put in the slide that it's the applications that need to evolve along with the infrastructure, right? You can't just rely on infrastructure to evolve and the application doesn't catch up, and then you're going to have this disparity that's going to be similar to how networking evolved, right, networking infrastructure evolved, but then it's very disconnected from the applications. So in the same light, in this case, unless the applications evolve, you're not going to be able to get this whole next-gen architecture with it. That's where, for example, an elastic cloud storage, when we built it, we relied on the fact that we'll build the persistence, make the application development easier. So let's say any application, this media application that they're building, they get the persistence, they get the geo-replication, they get both those nice functionality by just writing to our APIs, and the fact that they're running on containers is a very good abstraction for them. So they can move their containers across data centers, they can bring it up in a completely different data center because they have a global namespace from the underlying infrastructure. They get that benefit from how ECS is giving them that advantage. So Kamal, one of the things that's just blown me away at this show is when you talk to the Docker people and people that have worked on some of these projects from the kind of ideation to delivery is ridiculously short periods of time. And you can say those plugins we had Patrick on from the technical staff, he's been with the company three months and he did a bunch working on these plugins. We had Guy from Lyft talk about what Docker allows them to do is when they hire a new developer, they spend the morning doing an introduction and the afternoon they're deploying. I mean they're just moving in there. How's a company like EMC look at this? Because I worked at EMC and we released cycles in 12 to 18 months kind of development and things are moving so fast. So I'm curious how that shift impacts what you're doing inside EMC. Sure, sure. So I think that's a great segue into the DevOps side of things, right? So we started this team called the EMC code which the specific charter that like the goal of the team is to help the DevOps get better, right? So not just for the EMC products but as you're seeing we're contributing back to some of the technologies available out there in the community. So one of the advantages by doing that is that with the wider community participation and pooling resources we get to get to the stage where we can go with three or six month release cycles for the software. So if you look at how ECS is involved, we went with ECS 1.0 in last EMC world. We came up with another major release in December, ECS 1.1 and now we're at ECS 2.0 this June, right? So if you look at it, every six months we're able to get to a stage where we are doing some major releases along the way and one of the key advantages for that is going to your example of get the developer on board in the morning and evening, here's your container, get loaded into your laptop, start coding away, right? That's one of the big advantages that they have that they can do their dev test and they can do their own prototyping and hacking away and with some of the things like how EMC is not just using the technology of what we're producing, it's almost like a dark foot of what our own technology is, so we try to back up to our ECS and we try to dark foot some of those technologies so that we try to test it before we give it out to the community. So that's where some of the EMC code aspects also come in. They build some cool applications that they try to dark foot and those applications like mosaic me are out there in the web with ECS backed up. Cool, last question I have for you is, when did you first hear about the containers and just paint for us that story as to when you started it to where we are today? Sure, I would say our journey in containers started somewhere in the December to January timeframe of 2013, 2014, right? So literally like six months before our ECS 1.0 release, we said, okay, here's the technology, should we place a bet? We had like a lot of discussions, a lot of engineering, discussions, a lot of back and forth. We found that like from not just from the engineers bottoms up, but from the management, there was a lot of buy-in, a lot of confidence building exercise that happened, which gave us the thing that like, okay, let's place the bet on this. It was pre-GA, as I said, but we went ahead with it. A lot of dev time went into building this fabric, building some of the underlying technology so that I could get those containers to production. We went over that hump, the big hump, it took us the initial six months to get there, and then we were at the line with ECS 1.0, where we said, okay, we have a production version of a cloud storage, stateful container application running and here it is, right? So we were there, but we didn't stop there because once you go to customer environments with like thousands of nodes, you have challenges, right? So over the next six months to this year, we had to evolve some of the underlying technologies on how they fabric, the deployment aspects and fast forward one year, we had a version which is much more scalable, thousands and thousands of nodes, better buy the storage, and I think we are ready for prime time on this. All right, so before I let you off the hook, what were the two or three things that pushed you over the edge? That was a really quick time frame on which you made that bet, and you guys put a tremendous amount of investment right behind that bet. So what were the two or three things that as a team, you said, this is why we're going to place this bet? Right, so I think one of the first things was, we knew we had some of the key aspects in the IP to build this geo-related and strongly consistent architecture, which was very different than some of the technologies out there in the industry. So we knew we had the IP for this. Now taking this IP and taking this technology and getting it to the customer, right? Packaging it, making it deployable, making it in a completely distributed environment, not just from the software container perspective, software aspects of the storage, but getting it into production with the customer environments and commodity. I think that was the second aspect which pushed us into getting into this container world, right? And once we, and we've not done this, this is not the first time we've been doing this, right? With our previous generation of object storage, we were running exabyte level storage with 800 drags, multiple thousands of nodes at scale. So having some of those learnings, we knew we had to get to the scale and exceed that scale in a very short time frame. So I think we have some very aggressive targets for this year to get to this scale, and that's where it's pushed us and being the leading edge of technology at this elastic load storage. Exciting times, very exciting times. I'm looking forward to it. Big company at a little show that's grown crazy, so it must be fun to be at this kind of place inside the EMC world where it's a really super dynamic, kind of place. Absolutely, very excited to be here. Kamal Srinivasan, principal product manager EMC Advanced Software, thanks for spending a few minutes with us here at theCUBE. Good luck the rest of the show, sounds like you've been busy with sessions and trying not to get in trouble with the fire marshal. We don't like that. We like to keep the fire marshal on good terms. So I'm Jeff Rick, he's Stu Miniman. We're at TalkerCon 2015 in downtown San Francisco. We'll be back with our next segment after this short break. Thanks for watching. Thank you.