Live from Las Vegas, it's theCUBE. Covering Adobe Summit 2019. Brought to you by Adobe. Hey welcome back, everyone. Cube live coverage here in Las Vegas for Adobe Summit 2019. I'm John Fir with Jeff Frick. We're here with a Cube alumni. Been at Adobe for three years, Anjul Bhambhri, vice president of Platform Engineering at Adobe. Great to see you, thanks for coming by. Thank you, thank you, John and Jeff. All right, let's talk engineering. That was your line on the keynote. Great keynote today, by the way. Thank you. I was super impressed with the content. I'm watching that slide you were presenting, I'm like, we're at a cloud company. I'm feeling like I'm at Amazon re:Invent here. You guys built a really cool platform. Take us through, this was your mission. That's true, so-- So take us through your journey. Sure, sure. How'd we get here? How did Adobe get this beautiful platform? So, you know we've been at it for a few years and as we've seen CIOs and CMOs, right? That their focus is to really deliver delightful experiences to their customers. And not just once, but throughout the journey of the customer, right? Delight your customer every step of the way is what you'll hear from Adobe, from our customers. And we are really helping them to do that. And obviously in order to do that, there is, and as well you know, that data is behind everything to do with experiences as well. There is a lot of interactional data, and bringing it all together to really understand that holistic view of the customer is super important. And as you build the holistic view of the customer, it's not that you just build it once and you forget about it, right? You have to build this in real time. Because the interactions that customers are having with brands are through web, through mobile devices, through the apps that they are using of those brands. And the businesses have to understand that whole journey of the customer. And understand what their preferences are, right? What they like, what they don't like. And be able to, keeping that context, really during their journey whether they're coming to their website for the first time, or they are a repeat customer, be able to give them the right experience at every touchpoint. And that's where you need all of this data, which is a lot of data. (laughs) So you know, we've been on this big data journey, and me personally, even for a long time. But the scale that I've seen here, I had not seen before. Our IBM conversation, when you were at IBM prior, from Hadoop World, you had your eye on this big data trend. Now at Adobe when you have real data coming in with applications out in the market place, to put a platform together is a hard task. And I want to ask you a specific question around that. Looking at the architecture slide you have an analytics cloud, an ad cloud, a marketing cloud, and a commerce cloud. They all have markets that they have to address and be highly effective almost as a pure place stand-alone. But now integrating across each other now with a journey that you guys have put together is difficult. I know that from a computer science background. How'd you guys look at that architecturally, what were some of the guiding principles around building that platform? 'Cause you don't want to compromise the capabilities of those functional elements. Surely. So you decompose 'em, I get that. How did you put it all together? What was the key guiding principle around it? Yeah, so that's a really good question because Adobe has been delivering applications, right? Like you said, whether it's around analytics, or marketing cloud, or advertising. And now we've obviously just acquired the commerce cloud. And when you look at the common stuff around all of this, it's data, right? Data being captured through different channels, data that needs to be curated, having a common data dictionary so that things mean the same even though they are captured through different channels. So gathering this data, curating this data, organizing it for that holistic view of the customer, organizing it so that you can do BI and reporting on that data, is all something that we pull together in the platform layer. Now it becomes that whether it is you are doing analytics on this, right? Which could be your BI and reporting, or you are doing AI and ML on this to do your next best action, or you're targeting these customers with personalized content, you are doing it on a single version of the truth, which is the real time customer profile that powers all of these different clouds. Right. >> Right? So that it's not like when you do reporting you have one view of a customer, but when you are trying to show them personalized content, half the view is lost because the data was siloed. So we have gone past all of that. There's no data silos now, right? So realtime customer profile is literally being updated all the time, that's the key ingredient. That's the exciting part about. Across it all. >> Yes. So just curious, kind of philosophically in execution, 'cause like you said, you've been in this space for a long time. And one of the jokes I love to share is that we used to make decisions based on a sampling of something that happened in the past. Now, we can make decisions based on all of the data that's happening now. Yeah. But at the same time your challenge is that the sources are changing all the time, the speed of the input is changing all the time, and the expected return on your reaction is shortening all the time. Totally. So, from just a data professional, now am sure it's super exciting and super scary to move that paradigm shift to, you got to deliver the right thing right now. Totally. And you know, one of the key things here is that as all of this data is being gathered, right? Obviously this data has to be gathered where these events are occurring. So if you look at brands, their customers are global. They are transacting, browsing, whether it's on web, mobile devices, with that brand globally around the world. That means data has to be collected from these globally distributed edges. And it has to be brought in, processed in real time, building that profile. And as the data keeps coming, the profile is updated. And you can't have stale data in there, right? Because otherwise you are actioning based on something that happened five minutes ago. You know how we've seen that you buy something and you're still getting ads of that same product that you buy even a day or two days later. Right, I already bought the tent, I don't need anymore tent ads. Yes, that's because that brand has a stale profile of you, right? But if they had the real time customer profile, then there's no way that they would be delivering or actioning based on that stale information. So just like the data's been gathered from edges, even when we have to deliver the experiences, this is where edge computing comes into the picture. So we are also taking. So when you look at the whole architecture of the platform, yes, it's based on the cloud, and it's a big data stack, it's completely a SaaS offering, but there is also a big edge computing part of the platform, which is where all the hot data is collected, processed and actioned. And to your point, as we build say, predictive models on next best action on the data that's on the cloud, the scoring of the models has to happen on the edges where the events are occurring. So this is a complicated engineering problem but that's why I guess we love it. (giggles) Big smile. So the data's critical. Let's talk about how Adobe's changed over the past few years. Because you guys did cloud and I heard the nuance, I heard it in the keynote, reading through the lines is that it's hard to get data right at the beginning. Yeah. You get cloud right, now you got data right. Take us through that point because this is where I think the key to success is. How to make that data work. Because if you're going to have open APIs and open data, integrity of that data, the right database, whether it's a time-series or graph database, a lot of different applications might choose certain technologies. Yes. You have to deal with that. How important is the architecture on that? So that's why, that's a great question. You know, from a platform standpoint, our goal is that we have to be able to answer the questions with the right latency or speed as well as relevancy, right? So when we talk real time it's about... Latencies, you know when you talk to engineers they'll only talk latency, but it's not that. It's latency and relevancy. So in order to, depending on like if it's more like BI, or reporting kind of questions or queries, you need to organize the data a certain way. For single look-ups of customers, you have to organize the data differently. And that's where our IP comes into the picture that how do we partition and organize this data to meet the needs of both operational as well as the more analytical kind of work loads? So we support both, and to your point also that you know, when we need a SQL database versus a no SQL database, or a graph database, I mean those are choices we make. But on top we are providing APIs. So we are abstracting all of that from the user. And how, where we direct their question, that's all our IP. But their applications are not going to break because they're writing to the APIs. So, as technologies advance underneath, we make those right choices, but again, so that they are getting the right latency and relevancy. So in the cloud game we used to talk about this when you were on theCUBE we were at IBM, the DevOps movement was full tilt. And they used the term infrastructure as code. So you're kind of getting at, I want to get your reaction to this is that, if applications and workloads are the use cases are going to determine the data structures, data architecture and latency relevance equation, isn't then there's a new kind of infrastructures code emerging? Is it data is code? Or maybe ask this. Should the workloads dictate what type of data diversity and latency relevance is needed? Or does that come from the network? Again the question, is the workloads are kind of in charge I guess, is what I'm trying to get at, so this isn't-- Yeah, I would say that as a platform you have to support all of these workloads, right? So which means that from an architecture standpoint, we have to make sure that whether it's an analytical kind of a question or workload like BI reporting, whether it is more like an operational kind of a question around that, you want to just do a quick question around what did this customer buy? Or what transaction happened? The underneath data structures and databases, we have to pick the right ones so that we are able to support both. Because you can't-- I guess the expectations of the workload. It is, yeah. So if you're running commerce, latency and relevance, low latency is going to be in the milliseconds or-- Correct. And relevance does have a high bar there too. Exactly. An analytics query for a BI tool might be-- May be different. >> Three seconds. Yeah. So again, this is a huge delta in terms of capabilities, having that all happen on the fly is hard. Yes. >> How do you guys do that? What's the secret sauce? Yeah, so that's the underlying technology, that the way we are building that is so that you can support both of those. And we, with the customers we are sticking to. That if they want SQL access to the data, they're getting that SQL access. Now depending on the kind of queries, whether they fall as BI and reporting, or more like transactional kind of things in nature, those are the right technical choices that we are making behind the scenes, so that to the user those are not apparent. Because they can really focus on the insights that they are getting and really making decisions based on that insight, and not get caught into how to build all of these different pieces so that they can support both of these workloads. The other thing is that, a lot of the time that has been spent in IT has been to figure out all of this so that the CIO can support the line of business like the CMO. Now, by Adobe taking care of all this, all this heavy lifting that IT had to do, I think that IT will be able to meet the requirements of the line of business much faster and there's going to be the agility that is needed to support the business. I think that's really our goal in how we support the CIOs so that they don't worry about all this technology, all the data management, how to collect all this data from globally distributed edges. I mean, that's the partnership that we are building with the CIOs, so that we help them in their journey of really helping their line of business deliver the best experiences. Anjul, great to see you having so much fun at Adobe. Thank you What's it like there? Tell us, what's it like working at Adobe? You've got a platform, certainly there's a lot of hard problems to solve. Yeah. So you got that on the engineering side. Tell us what the culture's like there? Adobe is a fantastic company. I mean I just love every bit. Every minute that I've spent here is fantastic. It's great people, open culture, open to new ideas, and I guess all the creative cloud has brought this creativity which is infused in people. So it's just been a blast. And people recognize the importance of how data is so critical to delivering those delightful experiences. And it's very rewarding to just see how focused everybody is in the company, to really help businesses delight their customers. So it's very rewarding. And the ecosystem is great. Yes. How about the developer ecosystem, what's your reaction to that? Are they-- I mean Adobe IO is, I don't know are... Yeah, so if you think of all the creators that work with Adobe products and build their applications, I mean the ecosystem is very rich. So combine creatives and the data and IT, I mean, that's fantastic. So we should call this marketing native, like cloud native? You have a confluence of developers. Developers, creators... And creatives coming together. And we call them the-- Dogs and cats living together. I mean this is like-- We call them the experience makers, right? So we are really bringing experience makers, developers, data scientists, all together. It's a whole new level for Adobe. It's a whole new level. It's fantastic. Anjul, thanks for coming on and sharing your insights. This is theCUBE coverage live here at Adobe Summit in Las Vegas. I'm John Fir, with Jeff Frick. Stay with us, we're here for two days. We're on day one of wall-to-wall coverage at Adobe Summit. We'll be right back. (upbeat techno music)