 Live from New York, it's theCUBE. Covering the IBM Machine Learning Launch Event brought to you by IBM. Now here are your hosts, Dave Vellante and Stu Miniman. Welcome back to New York City, everybody. We're here at the Waldorf Historia covering the IBM Machine Learning Launch Event, bringing machine learning to the IBM Z. Brian Smith is here, he's the Vice President of R&D and the CTO of Rocket Software, powering the path to digital transformation, Brian. Welcome to theCUBE, thanks for coming on. Thanks for having me. So Rocket Software, a wall damn mass based, you know, close to where we are, but a lot of people don't know about Rocket. So pretty large company, give us the background. Yeah, so I've been around for, this will be our 27th year, a private company. We've been a partner of IBMs for the last 23 years. Almost all of that is in the mainframe space. We're focused on the mainframe space, I'll say. Around 1,300 employees, we call ourselves Rocketeers, spread around the world. We're really an R&D focused company. More than half the company is engineering and it's spread across the world on every continent in most major countries. Okay, and so you're essentially OEMing your tools, as it were, is that right? And no direct sales for us? About half, different lenses to look at this, about half of our go-to-market is through IBM, with IBM labeled IBM branded products. So we've always been, for the set of projects, we've always been R&D behind the products. The partnership though has really grown. It's more than just an R&D partnership now. Now we're doing co-marketing, we're even doing some joint selling to serve IBM mainframe customers. So the partnership has really grown over these last 23 years from just being the guys who write the code to doing much more. Okay, so how do you fit in this announcement? Machine learning on Z, where does Rocket fit? Yes, so part of what we announced with today is a very important piece of technology that we develop, we call data virtualization. Data virtualization is really enabling customers to open up their mainframe to allow the data to be used in ways that it was never designed to be used. So you might have these data structures that were designed 10, 20, even 30 years ago that were designed for a very specific application, but today they want to use it in a very different way and so the traditional path is to take that data and copy it to ETL, it's someplace else they can get some new use or to build some new application. What data virtualization allows you to do is to leave that data in place but access it using APIs that developers want to use today. They want to use JSON access for example or they want to use SQL access, but they want to be able to do things like join across IMS, DB2 and VSAM all with a single query using an SQL statement. So we can do that with relational databases and non-relational databases. So it gives us out of this mode of having to copy data into some other data store through this ETL process, access the data in place, we call it moving the applications or the analytics to the data versus moving the data to the analytics or to the applications. Okay, so in this specific case, and I've said several times today as Stu has heard me, two years ago IBM had a big theme around the Z13 bringing analytics and transactions together this sort of extends that. So great, I've got this transaction data that lives behind a firewall somewhere. So why the mainframe? Why now? Well, I would pull back to obviously we're seeing more companies and organizations wanting to move applications and analytics closer to the data. And the data in many of these large companies that core business critical data is on the mainframe. And so being able to do more real-time analytics without having to look at old data is really important. There's this term data gravity. And so I love the visual that presents in my mind that you have these different masses, these different planets if you will. And the biggest massivist planet in that solar system really is the data. And so it's pulling these smaller satellites if you will into this planet or the star by way of gravity because data is, data is a new currency, data is what the companies are running on. And so we're helping in this announcement with being able to unlock and open up all mainframe data sources, even some non-mainframe data sources and using things like Spark that's running on the platform, that's running on ZOS to access that data directly without having to, to write any special programming or any social code to get to all their data. And the preferred place to run all that data is on the mainframe if you're obviously if you're a mainframe customer. One of the questions I guess people have is, okay, I get that, it's the transaction data that I'm getting access to. But if I'm bringing transaction and analytic data together a lot of times that analytic data might be in social media, it might be somewhere else, not on the mainframe. How do you envision customers sort of dealing with that? Do you have tooling to help them do that? We do. So, so this differentiation solution that I'm talking about is one that is mainframe resident but it can also access other data sources. It can access DB2 on Linux, Unix, Windows, it can access Informix, it can access Cloudant, it can access Hadoop through IBM's big insights. Other feeds like Twitter, like other social media, it can pull that in. In the case where you would want to do that is where you're trying to take that data and integrate it with a massive amount of mainframe data, it's going to be much more highly performant by pulling this other small amount of data into next to that core business data. So I get the performance and I get the security of the mainframe. I like those two things, but what about the economics? Yep, so a couple of things. One, IBM when they ported Spark 2 ZOS, they did it the right way. They leveraged the architecture. There wasn't just a simple port of recompiling a bunch of open source code from Apache. It was rewriting it to be highly performant on the Z architecture, taking advantage of specialty engines. We've done the same with the data virtualization component that goes along with that Spark on ZOS offering that also leverages the architecture. We actually have different binaries that we load depending on which architecture of the machine that we're running on, whether it be a Z9, an EC12, or the big granddaddy of a Z13. Can you speak to the developers? I think about, you're talking about all this mobile and Spark and everything like that. There's gotta be certain developers that are like, oh my gosh, there's mainframe stuff. I don't know anything about that. How do you help bridge that gap between where it lives and the tools that they're using? Yeah, the best example is talking about embracing this API economy. And so developers really don't care where the stuff is at, right? They just want it to be easy to get to. They don't want to have to code up some specific interface or language to get to different types of data, right? And so IBM's done a great job with ZOS Connect in opening up the mainframe to the API economy with RESTful interfaces. And so with ZOS Connect combined with Rocket Data Virtualization, you can come through that ZOS Connect same path using all those same RESTful interfaces, pushing those APIs out to tools like Swagger, which the developers want to use. And not only can you get to the applications through ZOS Connect, but we're a service provider to ZOS Connect, allowing them to also get to every piece of data using those same RESTful APIs. If I heard you correctly then, developer doesn't need to even worry about that it's on mainframe or speak mainframe or anything like that, right? The goals that they never do, that they simply see in their tool set, again like Swagger, that they have data as well as different services that they can invoke using these very straightforward, simple RESTful APIs. Yeah, can you speak to kind of the customers you talk to? You know, there's certain people out in the industry. I've had this conversation for a few years at IBM shows is there's some part of the market there like, oh well the mainframe is this dusty old box sitting in a corner with nothing new. And my experience has been the containers and cool streaming and everything like that. Oh well, mainframe did virtualization and Linux and all these things really early decades ago and is keeping up with a lot of these trends with these new type of technologies. What do you find in kind of the customers that how much are they driving forward on new technologies, looking for that new technologies and being able to leverage the assets that they have? You asked a lot of questions there. So the types of customers certainly financial insurance are the big two but that doesn't mean that we're limited and not going after retail and helping governments and manufacturing customers as well. So what I find is talking with them that there's kind of the folks who get it and the folks who don't. And the folks who get it are the ones who are saying, well I want to be able to embrace these new technologies and they're taking things like open source, they're looking at Spark for example, they're looking at Anaconda. So last week we just announced at the Anaconda conference we step on stage with Continuum, IBM and We Rocket stuff. They're talking about this partnership that we formed to create this ecosystem because the development world changes very, very rapidly. For a while, all the rage was JDBC or all the rage was component broker. And so today it's sparking Anaconda are really in the forefront of a developer's minds. And so we're constantly moving to keep up with developers because that's where the action is happening. Again, they don't care where the data is housed as long as you can open that up. We've been playing with this concept that came up from some research firm called Two Speed IT where you have maybe your core business that has been running for years and it's designed to really be slow moving, very high quality, it keeps everything running today but they want to embrace some of their new technologies. They want to be able to roll out a brand new app and they want to be able to update that's multiple times a week. And so this Two Speed IT says, well you kind of break them off into two separate teams. You don't have to take your existing infrastructure team and say, well yeah, you must embrace every agile and every DevOps type of methodology. What we're seeing customers be successful with is this kind of Two Speed IT where you can fraction these two and now you need to create some nice integration between those two teams. So things like data virtualization really help with that. It opens up and allows the development teams to very quickly access those assets on the mainframe in this case while allowing those developers to very quickly crank out an application where quality is not that important, where being very quick to respond and doing lots of A-B testing with customers is really critical. Yeah, Waterfall still has its place. As a company that predominantly or maybe even exclusively is involved in mainframe, I'm struck by, it must have been 2008, 2009, Paul Moritz comes in and he says, VMware, our vision is to build the software mainframe. And of course, the world said, ah, that's great, mainframe's dead and we've been hearing that forever. And in many respects, credit to VMware, they built sort of a form of software mainframe, but now you hear a lot of talks about going back to bare metal. You don't hear that talk on the mainframe. Everything's virtualized, right? So it's kind of interesting to see and then IBM uses the language of private cloud and mainframe is joking, the original private cloud. So my question is, your strategy as a company has been always focused on the mainframe and going forward, I presume it's going to continue to do that. What's your outlook for that platform? Yeah, so we're not exclusively by the mainframe, by the way. Okay, so. We're not, we have a good mix. Okay, so it's overstating that then, it's half and half or whatever, you don't talk about it because you're a private company. Maybe a little more than half is. Okay, it's significant. It is significant. We've got a large proportion of the company on mainframe, ZLS. Right, so we're bullish on the mainframe. We continue to invest more every year. We invest, we increase our investments every year and so in a software company, your investment is primarily people. We increase that by double digits every year. We have license revenue increases in the double digits every year. I don't know many other mainframe based software companies that have that. But I think that that comes back to the partnership that we have with IBM because we are more than just a technology partner. We work on strategic projects with IBM. We'll oftentimes stand up and say, Rocket is a strategic partner that works with us on hard problems solving customers' issues every day. So we're bullish, we're investing more all the time. We're not backing away, we're not decreasing our interest or our bets on the mainframe. If anything, we're increasing them at a faster rate than we have in the fast- Well, this trend of bringing analytics and transactions together is a huge mega trend. I mean, why not do it on the mainframe? If the economics are there, which you're arguing that in many use cases they are because of the value component as well, then the future looks pretty reasonable, wouldn't you say? I'd say it's very bright. At the Anaconda conference last week, I was coming up with an analogy for these folks. Just a bunch of data scientists, right? And during most of the breaks and the receptions, they were just asking questions, well, what is the mainframe? I didn't know that we still had them and what do they do? So it was fun to educate them on that. But I was trying to show them an analogy with data warehousing where, say the mid-90s, it was perfectly acceptable to have a separate data warehouse separate from your transaction system. You would copy all this data over into the data warehouse. That was the model, right? And then slowly, it became more important that the analytics of the BI against a data warehouse was looking at more real-time data, right? So then became more efficiencies and well, how do we replicate this faster? And how do we get closer to not looking at weak old data, but day old data? And so I explained that to them and said, the days of being able to do analytics against old data that's copied are going away. ETL, we also mostly say that ETL is dead. ETL's future is very bleak. There's no place for it. It had its time, but now it's done because with data virtualization, you can access that data in place. And so I was telling these folks, as they're talking about these data scientists, as they're talking about how they look at their models, their first step is always ETL. And so I told them the story, ETL is dead and they just kind of looked at me kind of strangely. Well now the first step is load. Yes, there you go, right, yes, load it in there. But having access from these platforms directly to that data, you don't have to worry about any type of a delay. So what you described though was still a common architecture where you've got, let's say a Z mainframe, it's got an InfiniBand pipe to some Exadata warehouse or something like that. And so IBM's vision was okay, we can collapse that, we can simplify that, consolidate it. SAP with HANA has a similar vision, we can do that, I'm sure Oracle's got their vision. What gives you confidence in IBM's approach and legs going forward? Probably due to the advances that we see in ZOS itself where handling mixed workloads, which it's been doing for many of the 50 years that it's been around, being able to prioritize different workloads, not only just at the CPU dispatching, but also at the memory usage, also at the IO all the way down through the channel to the actual device, you don't see other operating systems that have that level of granularity for managing mixed workloads. And the security component, that's what to me is unique about this so-called private cloud. And I'd say it was using that software mainframe example from VMware in the past, and it got a good portion of the way there, but it couldn't get that last mile, which is any workload, any application with the performance and security that you would expect, it's just never quite got there. So it's, I don't know if the pendulum is swinging, I'm not, that's the accurate way to say it, but it's certainly stabilized, would you say? There's certainly new eyes being opened every day to saying, wait a minute, I could do something different here. Muscle memory doesn't have to guide me in doing business the way I have been doing it before. And that's this muscle memory I'm talking about of this ETL piece. Right, well, and a large number of workloads in mainframe are running Linux, right? You got Anaconda, Spark, you know, all these modern tools. So the question you asked about developers was right on. If it's independent or transparent to developers, then who cares? So that's the key, right? That's the key lever this day and age is the developer community. That's right, given what they want. You know it well. They're the customers of the infrastructure that's being built. All right, Brian, we'll give you the last word, you know, bumper sticker on the event, rocket software, your partnership, whatever you choose. We're excited to be here. It's an exciting day to talk about machine learning on ZOS. You know, I say we're bullish on the mainframe. We are, we're especially bullish on ZOS and that's what this event today is all about. That's where the data is. That's where we need the analytics running. That's where we need the machine learning running. That's where we need to get the developers to access the data live. Excellent, Brian. Thanks very much for coming to theCUBE. Thank you. And keep right there, buddy. We'll be back with our next guest. This is theCUBE, we're live from New York City. Right back.