(upbeat trumpet music) >> From our studios in the heart of Silicon Valley, Palo Alto, California. This is a CUBE Conversation. >> Welcome, everybody, I'm Dave Vellante and we're here in our Palo Alto studios and we're going to talk about big data. For the last ten years, we've seen organizations come to the realization that data can be used to drive competitive advantage and so they dramatically lowered the cost of collecting data. We certainly saw this with Hadoop, but you know what data is plentiful, insights aren't. Infrastructure around big data is very challenging. I'm here with Kumar Sreekanti, co-founder and CEO of BlueData, and a long time friend of mine. Kumar, it's great to see you again. Thanks so much for coming to theCUBE. >> Thank you, Dave, thank you. Good to see you as well. >> We've had a number of conversations over the years, the Hadoop days, on theCUBE, you and I go way back, but I said up front, big data sounded so alluring, but it's very, very complex to get started and we're going to get into that. I want to talk about BlueData. Recently sold to company to HPE, congratulations. >> Thank you, thank you. >> It's fantastic. Go back, why did you start BlueData? >> When I started BlueData, prior to that I was at VMware and I had a great opportunity to be in the driving seat, working with many talented individuals, as well as with many customers and CIOs. I saw while VMware solved the problem of single instance of virtual machines and transform the data center, I see the new wave of distributed systems, vis-a-vis first example of that is Hadoop, were quite rigid. They were running on bare metal and they were not flexible. They were having, customers, a lot of issues, the ones that you just talked about. There's a new stack coming up everyday. They're running on bare metal. I can't run the production and the DevOps on the same systems. Whereas the cloud was making progress so we felt that there is an opportunity to build a Vmware-like platform that focuses on big data applications. This was back in 2013, right. That was the early genesis. We saw that data is here and data is the new oil as many people have said and the organizations have to figure out a way to harness the power of that and they need an invisible infrastructure. They need very innovative platforms. >> You know, it's funny. We see data as even more valuable than oil because you can only once. (Kumar laughs) You can use data many, many times. >> That's a very good one. >> Companies are beginning to realize that and so talk about the journey of big data. You're a product guy. You've built a lot of products, highly technical. You know a lot of people in the valley. You've built great teams. What was the journey like with BlueData? >> You know, a lot of people would like it to be a straight line from the starting to that point. (Dave laughs) It is not, it's fascinating. At the same time, a stressful, up and downs journey, but very fulfilling. A, this is probably one of the best products that I've built in my career. B, it actually solves a real problem to the customers and in the process you actually find a lot of satisfaction not only building a great product. It actually building the value for the customers. Journey has been very good. We were very blessed with extremely good advisors from the right beginning. We were really fortunate to have good investors and I was very, as you said, my knowledge and my familiarity in the valley, I was able to build a good team. Overall, an extremely good journey. It's putting a bow on the top, as you pointed out, the exit, but it's a good journey. There's a lot of nuance I learned in the process. I'm happy to share as we go through. >> Let's double-click on the problem. We talked a little bit about it. You referenced it. Everyday there's a new open source project coming out. There's The Scoop and The Hive and a new open open source database coming out. Practitioners are challenged. They don't have the skillsets. The Ubers and the Facebooks, they could probably figure it out and have the engineers to do it, but the average enterprise may not. Clearly complexity is the problem, but double-click on that and talk a little bit about, from your perspective, what that challenge is. >> That's a very good point. I think when we started the company, we exactly noticed that. There are companies that have the muscle to hire the set of engineers and solve the problem, vertically specific to their application or their use case, but the average, which is Fortune 500 companies, do not have that kind of engineering man power. Then I also call this day two operations. When you actually go back to Vmware or Windows, as soon as you buy the piece of software, next day it's operational and you know how to use it, but with these new stacks, by the time stack is installed, you already have a newer version. It's actually solutions-led meaning that you want to have a solution understanding, but you want to make the infrastructure invisible meaning, I want to create a cluster or I want to funnel the data. I don't want to think about those things. I just wanted to directly worry about what is my solution and I want BlueData to worry about creating me a cluster, automating it. It's automation, automation, automation, orchestration, orchestration, orchestration. >> Okay, so that's the general way in which you solve this problem. Automate, you got to take the humans out of the equation. Talk specifically about the BlueData architecture. What's the secret sauce behind it? >> We were very fortunate to see containers as the new lightweight virtual machines. We have taken an approach. There are certain applications, particularly stateful, need a different handling than cloud-native non-stateful applications so what we said was, in fact our architecture predates Kubernetes, so we built a bottoms-up, pure white-paper architecture that is geared towards big data, AIML applications. Now, actually, even HPC is starting to move into that direction. >> Well, tell me actually, talk a little bit about that in terms of the evolution of the types of workloads that we've seen. You know, it started all out, Hadoop was batch, and then very quickly that changed. Talk about that spectrum. >> It's actually when we started, the highest ask from the customers were Hadoop and batch processing, but everybody knew that was the beginning and with the streaming and the new streaming technologies, it's near realtime analytics and moving to now AIML applications like H2O and Cafe and now I'm seeing the customer's asking and say, I would like to have a single platform that actually runs all these applications to me. The way we built it, going back to your previous question, the architecture is, our goal is for you to be able to create these clusters and not worry about the copying the data, single copy of the data. We built a technology called DataTap which we talked about in the past and that allows you to have a single copy of the data and multiple applications to be able to access that. >> Now, HPC, you mentioned HPC. It used to be, maybe still is, this sort of crazy crowd. (laughter) You know, they do things differently and everybody bandwidth, bandwidth, bandwidth and very high-end performance. How do you see that fitting in? Do you see that going mainstream? >> I'm glad you pointed out because I'm not saying everything is moving over, but I am starting to see, in fact, I was in a conversation this morning with an HPC team and an HPC customer. They are seeing the value of the scale of distributed systems. HPC tend to be scale up and single high bandwidth. They are seeing the value of how can I actually bring these two pieces together? I would say it's in infancy. Don't take me to say, look how long Hadoop take, 10 years so it's probably going to take a longer time, but I can see enterprises thinking of a single unified platform that's probably driven by Kubernetes and have these applications instantiated, orchestrated, and automated on that type. >> Now, how about the cloud? Where does that fit? We often say in theCUBE that it's not Moore's Law anymore. The innovation cocktail is data, all this data that we've collected, applying machine intelligence, and then scaling with the cloud. Obviously cloud is hugely important. It gobbled up the whole Hadoop business, but where do you see it fitting? >> Cloud is a big elephant in the room. We all have to acknowledge. I think it provides significant advantages. I always used to say this, and I may have said this in my previous CUBE interviews, cloud is all about the innovation. The reason cloud got so much traction, is because if you compare the amount of innovation to on-prem, they were at least five years ahead of that. Even the BlueData technology that we brought to the barer, EMR on Amazon was in front of the data, but it was only available Amazon. It's what we call an opinionated stack. That means you are forced to use what they give you as opposed to, I want to bring my own piece of software. We see cloud, as well as on-prem pretty much homogenous. In fact, BlueData software runs both on-prem, on the cloud, in a hybrid fashion. Same software and you can bring your stack on the top of the BlueData. >> Okay, so hybrid was the next piece of it. >> What we see is cloud has, at least from the angle from my exposure, cloud is very useful for certain applications, especially what I'm seeing is, if you are collecting the large amounts of data on the cloud, I would rather run a batch processing and curate the data and bring the very important amount of data back into the on-prem and run some realtime. It's just one example. I see a balance between the two. I also see a lot of organizations still collecting terabits of data on-prem and they're not going to take terabits of data overnight to the cloud. We are seeing all the customers asking, we would like to see a hybrid solution. >> The reason I like the acquisition by HPE because not only is it a company started by a friend and someone that I respect and knows how to build solid technology that can last, but it's software. HPE, as a company, my view needs more software content. (Kumar laughs) Software's eating the world as Marc Andressen says. It would be great to see that software live as an independent entity. I'm sure decisions are still being made, but how do you see that playing out? What are the initial discussions like? What can you share with us? >> That's a very, very, well put there. Currently, the goal from my boss and the teams there is, we want to keep the BlueData software independent. It runs on all x86 hardware platforms and we want to drive the roadmap driven by the customer needs on the software like we want to run more HPC applications. Our roadmap will be driven by the customer needs and the change in the stack on the top, not by necessarily the hardware. >> Well, that fits with HPE's culture of always trying to give optionality and we've had this conversation many, many times with senior-level people like Antonio. It's very important that there's no lock-in, open mindset, and certainly HPE lives up to that. Thanks so much for coming-- >> You're welcome. Back into theCUBE. >> I appreciate you having me here as well. >> Your career has been amazing as we go back a long time. Wow. From hardware, software, all these-- >> Great technologies. (laughter) >> Yeah, solving hard problems and we look forward to tracking your career going forward. >> Thank you, thank you. Thanks so much. >> And thank you for watching, everybody. This is Dave Vellante from our Palo Alto Studios. We'll see ya next time. (upbeat trumpet music)