 Live from Las Vegas, Nevada, extracting the signal from the noise. It's theCUBE covering IBM Edge 2015, brought to you by IBM. Okay, hello everyone, welcome back to IBM Edge 2015. This is Silicon Angles theCUBE, our flagship program. We go out to the events and extract the signal noise. Live in Las Vegas, I'm John Furrier with my co-host, Duke Miniman, our next guest is Randy Furking, enterprise technical expert, Walmart. You're involved in all the action, Walmart, a lot of pressure, a lot of volume on retail transactions. Welcome to theCUBE. No pressure whatsoever. So you guys are a big IBM customer, and you got a great story, I want to dig into it a little bit. So first tell us a little bit about your role and some context to what's going on at Walmart around some of the challenges around scaling, large amount of transactions, low latency. What are some of the challenges out there? One, in selecting products to do that, and two, use cases that you're under pressure to deliver because your customers are always banging on walmart.com, making transactions on the retail front. What's happening? Right, while I hired into Walmart back in 2007, we averaged about 200 million transactions a day through our online system. We run on a Parallel Susplex ZOS running CICS. And in 2009, we introduced what's called web services using SOAP and SOA services. We now run about 500 million transactions a day, so we've more than doubled almost two and a half times the volume on the system. And with that, you have the uncertainty of transaction volume. So we have different holiday seasons where the volume increases, and we have no idea what that volume is going to be, right? It's kind of hard to predict what the customer base is going to do, whether it's from the .com, from a mobile app or multi-channel platforms. So, we'll start with an example. We had a customer-facing application about two years ago that needed a enterprise distributed caching solution for the next production rollout. They had purchased a product that was having performance and availability issues. So, we decided to design and write a cloud-based service running on ZOS Parallels Susplex and CICS. So we created a caching service, the caching as a service solution. Since putting that into production, this application or this customer hits it about 15 million times a day. That's over 10 and a half billion calls over the last two years with zero failures. Zero failures. Zero failures. That's active across two different data centers. So you wrote software for the cloud for ZOS. Yes. For Z systems. So talk about that dynamic. How many people, what was it like, and why? Well, first of all, Linux, whether it's X or Z, already has a good presence with cloud. What's missing in the industry is ZOS, the Parallels Susplex. So, my partner, my colleague, Rich Jackson, and I, we attended ZBLC where we worked with the council on cloud platform vendors, ZOS. And so what we wanted to do was look at how do we utilize the assets or the capabilities of ZOS and make the Susplex, instead of cloud-like, cloud. So we developed our own Cell Service Provisioning Portal. It's all written on CICS, all written on the mainframe. And our services, all of our cloud services again, are written CICS, ZOS, Parallels Susplex. So we designed and wrote those ourselves. So here's some color around the scale and the failure range. Zero failures. Yes. The numbers are pretty massive. Are you guys tweaking the system? Is it self-healing? I mean, what's going on with the system? Why no failures? And did, was there work involved? Was that just the system doing all the work? Right. Well, you make a good point. Had we written these services on, you know, the distributed platform, we would have the same challenges and the same failures that occur on that platform. Because we utilize Parallels Susplex, workload manager, Susplex distribution, we're able to take those unexpected workloads and manage those across the entire Plex. So what about the big flash mob that comes in on the retail trends? You guys, because you guys see very lumpy, you know, traffic, you know, the big shopping days come in. What's the shift and scale? And how does the system handle things of that nature? So back in 2010, we developed on the mainframe in CICS an enterprise or a application. Basically, it's a router, a browser that runs on the mainframe. And it averages about 30 million transactions a day in production. So it's not a real heavy application, but it gets a little bit of traffic. Well, last year on Black Friday, it scaled out to 190 million, six times the volume. So we didn't have to change anything within the configuration of the CICS servers or of the Susplex itself. That's what workload manager does. That's what the Susplex does is it manages that workload across the Plex. So no change is just automatically rolled over or rolled the scaled up and no problems. Yeah, so we also have a, as our caching solution, we have a mobile app for our global e-commerce and where we cache all of our HTML pages in our ZUS caching service. The average of about 15 million transactions a day going into the holidays, it increased to 25 million. On Black Friday and Cyber Monday, it increased to 80 million. We did nothing to change the environment. We did not add any more servers, did not add any more capacity. Here again, that's what the parallel Susplex does. About price-wise, you're paying out the nose for this. What's the price difference between the alternative solutions you add? So we did a cost study on our caching service on ZOS compared to comparable services on X and it was a five to one cost ratio in favor of ZOS. So less expensive and no failures? Yes. And the other one was five times more expensive with fail. With performance and availability issues. Hashtag fail, whatever they want to say. That's a good deal then. That's kind of a no-brainer, is it? Yeah, so Randy, in the industry today, there's so much discussion about scale out, distributed architecture. There's few companies that have the global presence that Walmart does. How do you architect for that kind of global market and at that scale they do? How do your folks stay up with the latest technologies and trends and move the IT forward? Well, here again, we work with ZBLC, which has been a big part of our success. So we're up to speed on the current technologies that the Parallel Susplex offers. Current storage technologies. Up the current availability of CICS and features within CICS. So with those, we just simply utilize what IBM's put together and we come up with scalable solutions. Yeah, if you look at the hyper scale, players out there, there's a lot of discussion that they really just build the application layer and the infrastructure kind of takes care of itself. Do you still have an infrastructure team that manages things? How do the infrastructure and the application sit together? So we do have an infrastructure team, we have storage teams, we have the middleware team as well, and you have your application teams. And if we all stay within our own silos, we're going to be very disconnected and have a lot of different issues. So it's very important that we communicate across those tiers and work with the application teams to see what the requirements are, see what the workload is expected to be, and then make sure that within the infrastructure and the middleware teams, the storage team that they're working together to make sure we have the right solution in place. We did the big Z launch in New York City and one of the things that impressed me was the mainframe's relevance because of the power now, all this kind of new capability, encryption, in-memory analytics. Is Walmart tapping into some of those new things and how do you take the success that you've had and replicate it, or is there other use cases in Walmart that you see potential of this? Because, I mean, the economics are compelling. Obviously, the scale-out is awesome. It seems like a really big deal. I mean, what's the internal use case? Is there just a one-off? Well, think about this. The mainframe is server, okay? Used to be called the Z Enterprise server. It's just a server, just like an other server. It's just a big server, and it's a reliable server, right? So with that, it has features like encryption with the crypto engines, crypto cards. And so we're starting to utilize the cryptography with NZ as services for non-Z platforms. Tell me about the database challenges because we do our own big data stuff in our company. I know other customers, they use a lot of open-source stuff. Is there a database selection and architecture that was failing? Why was the other guys failing? And why was Z successful? Because you guys wrote your own code so you cloud-enabled the solution. But yeah, it's a big server, so you got a database there. How does the database not fail at that kind of scale? Okay. Well, IBM provides on the mainframe a hierarchical database called IMS, relational database, DB2. And what my colleague and I decided to do was fill a void that was missing on ZOS and that was we created a key value database as a service, as a cloud service. And so with that, we've been able to utilize the simplicity of a key value database instead of putting all the burden on DB2 for relational. In doing so, we've been able to be very innovative. Our key value database is a service, rest-enabled only. You can use structured run, structured data. You can serve up and store GIF, JPEG, PDFs, videos up to two gigabyte object sizes. So we saw a need and we filled the gap. We just designed it. So you leverage it with the architecture of Z. Yes. It's not so much retrofit into DB2, so to speak, right? DB2 is relational, okay? And not everything is relational in nature. The cloud service that we created for a key value store was not to compete with or take away from DB2, not whatsoever. It's to take the burden off where people were putting non-relational stuff into DB2. Like on structured data and making all this social data, buying data, blues data. Exactly. So we're able to store, like I say, videos, video clips, product information, photographs, whatever it is that we want within a key value store, which is much quicker access than going through a relational or hierarchical system. So Randy, Walmart's well-known for really pushing the envelope and taking their suppliers and helping them to move forward where the industry needs to go. How are things going in your viewpoint? And you mentioned kind of the ZOS. Are there other things that you're looking for from the community or things that you think we need to move forward to help you run your business better? Well right now my focus, I can't speak to the rest of the industry, my focus is on the Paralysis plaques as a cloud platform provider. We work with IBM, the partnership with the team at Poughkeepsie and we work with ZBLC and other ZBLC customers and that's been extremely relevant and helpful to us in how we take our technology forward to seeing what the rest of the industry requires. So the Z thing that gets my attention is also encryption. Is security a challenge or is that not prior to that consideration to you guys? Security is a extremely important part of the ZOS platform. So we rely heavily on Rack F and we rely heavily on the encryption decryption services that are built into the product. On the, give us a little inside color if you can on Walmart. What's it like there, tech-wise? You guys are writing your own code, pretty innovative culture. Share the folks out there. What's it like working there? You have unlimited budget. I mean, I mean, what's happening? I'm talking about the limited budget part. One of the things that Walmart technology does is they do encourage innovation on all platforms, whether it's in applications, whether it's on the X platform or the Z platform, it's irrelevant. Walmart is about innovation because when you are innovative, you're able to do things for the customer and that's what it's all about. Okay, because we can write code all day long. We can do really cool, fun stuff and it's absolutely meaningless if it does not benefit our customer. And so Walmart does encourage- Get data driven culture, isn't it? I'm sorry? Data driven culture, Walmart? I mean, performance matters. There's so much transaction going on. If you miss a B, revenue's impacted. Yes, but it's all about data. I mean, without data, what are you doing, right? Exactly. And what's about the future? What do you see next coming? What's the next, what's your attention here at the show? What's getting your eyes popping out? Is there anything you see here that's good that's looking to bring into the fold? Well, like I said, with ZOS being kind of new in the game of cloud processing, the steps that are being taken with IBM right now are extremely beneficial. Mobile is just another channel, right? So I started in this business back in 1979 and everything was 3270 data stream. But ZOS has kept up and it's evolved. I mean, it went from MQ series and TCP sockets and, you know, SNA. And now with HTTP, with SOAP services and REST services, you can handle any type of channel into the server, the Z server. It's a real multi-purpose server. It's not just mainframe. It's got Linux, it's got a bunch of Ramis. But mobile is the new customer facing. It is the way that the customers get to their data, get to Walmart's products, get to our information, get to our prices, get to our offerings. It's all about mobile right now. So with mobile, with cloud, and then with that, of course, is analytics. So what can we do to better serve the customer with the mobile channel, right? With the cloud versatility. And so it's about analytics as well too. But one of the most important things that we cannot overlook is security. And that's where ZOS really outshines the rivals. Randy, I really appreciate you coming on. Surely Walmart, great brand. You guys are a leader. Final question, more of a personal one. Share it with the folks out there who are watching who's been in the business. Not as long as you are learning the ropes. You're a veteran. Great success with this project. I mean, no failures, scales, amazing. You guys wrote your own code. A lot of experience and being relevant is key. What advice would you share to folks out there who are looking to get to a point where, using open source, they're doing innovative, disruptive things. What's the secret to the success that you've had? Passion. If you're not passionate about what you're doing, you should do something else. When you do have a passion, you still need to be persistent and determined. I have failed more times than I've been successful at writing a piece of code to do something. But if you're persistent and determined, you stay with it and you have that passion, you will eventually become very successful at what you do. And if you're just trying to be mediocre about it, if you're just going along with the trend, I don't follow trends. I want to set the trends. I want to take technology further. Yeah, so you got a little R&D, you got a little fail points. Don't go for the big fails. Go for, yeah, go bigger, go home, as they say. But that could be dangerous. If you go big and fail, is there a mix, is there a balance between risk reward? What's your formula for risk reward? Well, if we didn't take risk, we wouldn't be flying. We wouldn't be to the moon. We wouldn't have a lot of the capabilities that we have. Now there's, of course, reckless risk, but there's also too calculated risk. So if you look at things from a simplistic manner, test it out, you look at all the different possibilities, and then when you take that, and you work through those and you're successful, then you can scale that out with confidence. Randy, thanks for coming on theCUBE. Really appreciate it. Randy from Walmart, we are here, live in theCUBE, sharing some great insight, some successes and experience inside theCUBE. We'll be right back with more after this short break.