 Hi everybody, we're back. This is Dave Vellante of Wikibon.org and this is theCUBE, where we go to the event from the noise. SiliconANGLE's been doing theCUBE since early in 2010 and we're at the MongoDB Days Conference in New York City. We're at the Marriott Marquis. Of course, we had a big talk in the last two days on, you know, Bernanke says the economy's doing but the market drops 500 points. But up a little bit today, but it's kind of struggling. People are really watching to see what kind of Friday this is and you know, everybody's wringing their hands, watching the bond market. You know, is this a good correction or the bad sign of things to come? But it's all good here. We're talking about, you know, we're talking about fations in database technology or really leading the charge with new applications. And as you know, in SiliconANGLE, Wikibon, we've been covering this whole big data theme for quite some time. All the trends, I'm here with Jeff Kelly who's Wikibon's lead big data analyst. Jagesh Saheba is here. He is the chief architect at the ADP Innovation Labs. He's a practitioner and he's going to share with us some of the hands-on knowledge he has around building apps, modern day apps, using MongoDB. Jagesh, welcome to theCUBE. Good to be here. So tell us a little bit about your role as chief architect and generally and specifically a little bit about ADP Innovation Labs. So I'm the chief architect for ADP Innovation Lab. The lab's mission is very simple. We use technology to create amazing products for our clients. And this product is that most important resource. So ADP is one of the largest HCM service product provider in the world. Worldwide presence. Worldwide presence. And the Innovation Lab, we incubate products, technologies and mobile is one of the technology we introduced a few years. And it has been quite successful and MongoDB is one of that offering. Mobile has just taken the world by storm. We were at the vlog and that's all. Anybody was talking about, there's big discussion about is the web getting faster, works are getting faster and the like. But you've got now added complexity. The cloud, social data. It's like two steps forward once. And mobile changes everything. So talk about some of the challenges from an architecture standpoint and how you guys are addressing them. So mobile did change everything, right? So one of the thing that we saw happening with the newer tech mobile tsunami building is that the consumers are demanding experience. This access, does anywhere, any type of their own device to work use it. Devices are personal. So from an enterprise application perspective, we wanted to create an architecture and product that addressed a lot of the demand consumer texture, those demands. So talk about some of this that you, particularly from the consumers standpoint because as a consumer, performance is everything, right? When I go on my mobile, I get worse performance. And so how do you, in this day and age, figure out the network, database, is it the storage? Is that a challenge for you in that? So you're right. The mobile introduced the fact that they weren't available any time, anywhere. Even on a low latency, even with the latency of the network, the cellular network, they weren't as applications in real time with a nice experience. We set out to creature a web scale group that can millions of you. So we looked at all the components of the front end, high performing, extreme scale to millions of users. And we accounted for latency as we, for ADP services, through a single application on the mobile platform. Merits of systems behind the scene, treat your data, we're building the cache. So what can we do to improve the production? We were prefetching the patient, you would come use the application and access your data at certain point in time. So this is where the behavior analytics, using big data analytics to have that feedback cycle so we can prefetch data from our backend systems, make it available for you or with a web socket technology we could even push it to your device and have it ready for you by the time you're ready to consume that information. Using analytics and machines to make decisions about where to put data so that you can have the user experience be more pleasant. Absolutely, we capture every impression with the goal of understanding how the product is being utilized, how an end user typically interacts with the product and to do predictive caching, right? So one of the example that I use is we know that you come every other week because that coincides with your pay cycle and you review your pay statement around nine a.m. in the morning. So I can go prefetch that data and make it available to the application so the application doesn't incur that latency of calling our payroll engine but I can prefetch that data, make it available at the application to your cache and if you use our native application I can even push that data to you and make it available, have it ready for you on the device itself. Well that's a fantastic example of the way big data is impacting product development. We were at the Excel Partners Stanford event a couple months back and talked to a lot of startups, companies like Prezi for instance who maybe you don't consider a big data company but they're using data to feed the development of their applications, to understand how people are using it and developing new services in their applications to make it as you say the best experience possible for the user. But so let's dig into the data itself a little bit. So obviously we're here at MongoDB Days. So data these days comes in all shapes and sizes and flavors, structured, multi-structured, unstructured. So I know you're using MongoDB as one of the technologies kind of under the covers to support your application development. Talk a little bit about the data sources, where you're getting this data from, the structure of the data and how Mongo and perhaps some other technologies are you helping you or making it possible for you to bring in all this data that's not in neat rows and columns. I think that technology such as MongoDB and NoSQL technologies play a big part into how we deal with data. So for ADP we interact with the numerous backend systems be your payroll engine, benefits engine, your time and labor management systems. So you can see the variety of data that can come out of the systems and the mobile application has to consume that and present to the end user. So the variety of data is there. If you have a database technology that requires fixed schemas, right? And it requires, it handicaps you if you have to, if you continuously have to modify your schema and have the schema migration as you roll out additional functionality. So MongoDB and NoSQL technologies are allowing us to deal data in a very agnostic way. Essentially, we can take any JSON data type, any JSON document and put it into cache and make it available for application to consume. As we're creating a web scale application, so the volume of data is tremendous. As we scale our applications to millions of users, we need to deal with the high volume of data. So we covered the variety of data, the high volume of data, and MongoDB is going to hit the sweet spot to manage data complexity that we have at ADP. Yeah, help us put MongoDB into context in the larger world of NoSQL. There's all different flavors of NoSQL. You know, we hear about HBase and Cassandra and some other things. Where is that sweet spot for Mongo? Is it somewhere kind of in the, where scale and variety kind of converge versus purely a good scale out database versus purely good for multiple types of data? Where does MongoDB kind of fit in that larger NoSQL universe? I think what MongoDB does uniquely is, you know, people go to NoSQL solution for scalability, for performance scalability. What Mongo is able to do is deliver on those promises of NoSQL as well, but at the same time offer a rich data manipulation functionality that our application requires. Right? In order to, you know, we can, we have REST interface, JSON documents, but at times you need to manipulate that data. And Mongo provides the framework to do that manipulation. And I think that creates that sweet spot of I'm not losing all the functionality of a traditional relational data store. I'm keeping that while I'm getting the scalability and the performance of a NoSQL data store. So translate that, if you could, into kind of making the business case to let's say a CEO or someone who's not in the IT world who doesn't necessarily understand NoSQL, doesn't even interested in what NoSQL is or structured or unstructured. They just want to know, what is the business case for investing in a new technology when, you know, I'm a CEO and we use Oracle, we use traditional databases. Why are we looking at some open source technology in this new technology? How do you, how would you articulate that business case to somebody who's not, you know, deep in the weeds into the tech itself? Right. So in terms of a business case for a technology like MongoDB and NoSQL, it's very simple, right? We are creating systems at a web scale. What typically happens with the traditional data stores is that you scale up. As your demand increases, you buy bigger and bigger machines to meet the demand. And so there are two curves. One curve is the cost curve, which exponentially goes up as you buy bigger and bigger boxes to meet the demand. And then there's another curve, which is the performance that comes to time where the sheer size of the infrastructure, your performance degrades, right? You want to flatten out those two curves. So one of the biggest selling point of a technology such as MongoDB is you're going to flatten out both curves. So you want to get consistent performance as you scale out. At the same time, you don't want your cost curve to exponentially go high. You want to flatten it out and make it much more manageable from a cost perspective. So we run MongoDB on a vanilla hardware, vanilla VM infrastructure hardware, and we're able to scale this fairly easily. I think that's an amazing business case to tell your CEO that we're building a web-scale application that can scale to millions of users, yet the infrastructure costs are not going to go exponentially higher as we scale more and as we get bigger and bigger. Yeah, that certainly would resonate with the CEO. Well, so I want to push on that a little bit and get an architect's perspective because there's increasingly, I mean, we watch the hyperscale guys, right? And we've been saying now for years, if you want to know what's going to happen in the enterprise, look at what's happening in Google and Amazon and Facebook, and it's going to seep into the enterprise. And it's clearly happening. However, there's some discussion around within the hyperscale community now the massive amounts of capital investment that they're putting in there. Wall Street's putting pressure on guys like Facebook and Google for all the capex that they're spending. And they're starting to, I think, struggle with some of the complexity of scale-out. I mean, let's face it. Scale-up, shared something is conceptually anyway easier to manage, at least in theory. At some point, does that complexity of scale-out get so great that that cost curve reverses, or do you believe that the industry will continue to allow that, you know, flatten cost curve and that balance that you described to be achieved? I think that's the beauty of where a company like Tengen can come in and address the enterprise needs. Right? So to enterprise like ours, yes, we will scale out. The infrastructure will get larger. There'll be more number of virtual machines or machines participating in the delivery. So enterprise capability of what they're introducing and a lot of the features are inherent to this architecture. Being able to have less hands-on operations requirements, but being able to manage cluster automatically, being able to manage shard automatically, being able to balance automatically. If you add additional node, being able to adjust to that increased capacity, if you take out a node or a component fails, these are all inherently built into the architecture. So, and I see company Tengen addressing some of the enterprise need around security, management aspects, right, with the on-premise management capabilities which is introduced in the latest release. They're addressing some of these concerns. But I think in terms of choices that we have, scaling upwards is scaling out. I think the value proposition is very clear in terms of scaling out with a commodity hardware. So that's over, I mean, in your mind, right? That debate is over. In my mind, it is manageable and I think there are some, there are security is one of the areas of concern and I see the focus on those aspects to address those. So we talked about mobile complexity. Cloud, obviously, in the one hand, simplifies things because you get this awesome resources that you can utilize. But for a company like yours, where you're probably looking at combinations of cloud, not just pure public, you don't have a blank sheet of paper to start from. So you're probably doing some kind of hybrid. That adds to the complexity, doesn't it? Particularly with the security component. Can you talk about that a little bit? I think the complexity is the tool set or the technology has built-in capability to manage complexity. Is it all solved? Probably not, right? I think the complexity for a company like ours comes in is when you introduce new technology into a stack, right? And the acceptance of it and the putting operational procedures around it, those are the complexity, but I don't think they're necessarily technology complexity much more like procedural complexity. People in process. People in process, right? And we're getting to that stage where within ADP, we see MongoDB as a service, as a platform being offered to application to make that adoption easier to various applications rather than they investing their time and effort building the infrastructure itself. How does DevOps play into your shop? In fact, we still do DevOps with MongoDB today. We don't have a dedicated DBA nor we have a dedicated operations person. So we believe in DevOps. The developers are able to manage and operate Mongo in production environment. We have automated a lot of numerous things through scripts. We have automated a lot of things. So it is not something that consumes a whole lot of their time. So I think DevOps gives us that opportunity to be in touch with your production system rather than throwing over the wall and somebody else manages your production system. So I wonder if we could just talk a little bit about the impact of Mongo on your plans for the future in terms of the other types of technologies you're working with. I mean, do you find that is Mongo going to be kind of the database of choice for building applications going forward? And is that impacting your investment in another, I'm sure you've got, Oracle Database is running somewhere. Is that going to impact, do you see this becoming the NoSQL world Mongo specifically but NoSQL in general becoming more and more critical to your infrastructure and potentially pushing out some of those relational, traditional, relational technologies that are in your environment currently? Pushing out is probably a harsher term. I think there is a need and there is a place to use technology such as this. And there is a need and a place to use a relational technology if I'm writing a payroll engine from scratch, right? I would probably use a traditional database technologies. But if you're creating a massively scalable presentation engine like mobile or even a portal system, right? This technology makes a lot of sense, right? So I see coexistence and the time will tell in terms of the feature sets and what comes within this database technologies. Today I see happy coexistence between the traditional relational database technology and NoSQL technologies like MongoDB. Jagesh, we're running out of time but the last word we'll give to you. What advice would you give fellow practitioners trying to move down this journey of new technology? NoSQL, open source, you talked about people and processes, maybe that's part of the advice but what advice would you give your peers? I think the key advice is that listen to your users. Everything we do at ADP Innovation Lab is user-centric, right? Listen to your user, understand what they're doing, understand where they're going, what devices they're using, how they want to consume your services and tailor your architecture for that need. And naturally, when you have a client base of ADP over 600,000 client base millions of users, you need to pick tool sets and technologies that can scale to that level, right? But it has to be driven by the need and I think that's the key advice that I would give. Listen to your user, build the architecture to deliver the ultimate performance, the ultimate simplicity and empower your user, inspire confidence in them and your application will succeed. Jagesh, thanks very much, great advice, love what you guys are doing, love the attention to the users, the mobile innovations, congratulations and thanks very much for coming on theCUBE. Thank you very much. Thank you. Keep it, keep it right there everybody. We're coming back next with Rack Space, big announcement today and this is theCUBE, SiliconANGLE's continuous production of MongoDB Days, we're here live in New York City, we'll be right back.