 Hi everybody, we're back. This is Dave Vellante of Wikibon.org. I'm here with my co-host, Jeff Kelly. This is theCUBE where we go out to the events, we extract the signal from the noise. We find the best guests and bring them to you and share with you, our audience, their knowledge. Andrew Ehrlichson is here. He's got an interesting role at 10Gen. He's vice president of education and of cloud services. Andrew, welcome to theCUBE. Thank you, thank you for having me here today. Yeah, you're welcome. So let's see, let's start with the day. How's it going for you? What are you hearing from customers and in both your roles, left-hand, right-hand? Yeah, no, the conferences are really awesome. We get to meet the community that's so critical to our success. I started off a day by giving our, building your first application talk, which is, Oh, that was you, okay. Yeah, that was me. And for me, doing anything live is always strange because I've actually, do most of the work with the online education and I have the benefit of hours and hours of editing for every moment I work. So people don't know really what it was like to see me live. Well, I mean, we are live at theCUBE. Exactly, exactly. So that was one of the first sessions that really caught my eye, it's a great title. So what do you tell people? How do you get started building your first Mongo app? Yeah, I mean, we start by polling the audience and figuring out what their experience levels are. And the amazing thing is that, we've had these conferences for a long time and at every single conference, a really large number of people are just getting started. Almost all of them have a lot of experience with relational databases. So that's a good reference point. And so we just try to highlight what the high level goals were for the product when it started and how do you build applications in it and not get stuck if you come from a relational world. So you have passion for education. We were talking off camera about that. So talk about that a little bit. Sure, yeah, I mean, I was at Stanford in the PhD program there and three years ago, four years ago now, Khan Academy popped up teaching all these kids basically middle school math. And from there, there was a couple of classes, the AI class, the machine learning class that launched at Stanford. And I actually took them at the time after I graduated, 100,000 students in these classes. And I had tried lots of these online ed things before through iTunes University and whatnot. I'd never managed to get through the classes. And I took those classes and I realized that they had really cracked the code on the essential elements to motivate people to get through these classes. And so I said, wow, this is amazing. This is really disruptive. And so when I was looking around for the next thing to do, at this point, Coursera, Udacity, and edX at all launched, and I knew the founders of TenGen, I approached them and I said, I want to do this. I want to do this for TenGen. I want to do this for MongoDB. I want to do these online classes with 10, 30, 40,000 people in the audience who are going to teach everyone about MongoDB. And prior to that, we had taught maybe 1,000 people in person since the inception of the company. And they thought, Andrew's obviously insane. So let's try it. Yeah, but I want to work here. So let's just indulge him and if it doesn't work, we'll put him on something else in six months and it'll be fine. Wait, what's the worst that can happen? What's the worst that can happen? Exactly, right, that's right. So we came in on that basis. Actually, I think they were so uncertain about it that my original offer letter, rather than saying vice president of education, just said vice president. And so people introduced me around the company. This is our new vice president. I don't know what it's going to be working on. So based on that, we approached these different companies that were really at the forefront of online ad and edX was receptive to working with us. This is the Harvard MIT consortium. So we got together with them. And actually, Nant Agawal is their director and he's the first PhD student of my advisor at Stanford and I was probably his last. And Nant is a famous dude. He's the head of the AI lab at MIT. I don't think he knew me, but I knew him. So I approached him and said, we want to use your software. He said, okay. So we launched the first classes and we had 30,000 people sign up for them. And it was amazing. It was just, it was, and to be honest, what we're doing, the techniques are not that original. They're essentially very close copies of what's going on in higher education, but very few companies have used it for workforce development and training. And so far you've had the numbers, I believe 70,000 enrollment? Yeah, yeah, we've had 70,000 plus people, registrations. Some people registered more than one class since the inception about a year ago. One year. Yeah, so. You talked about the sort of light bulb went off when you were taking some of these online courses. What was it about them that made them so compelling and how did you adopt that for your courses? That's a great question. That's a great question. You know, when you look at these courses, like the way they were efficiently offered, like there will be a Stanford course on learning iOS programming. And so you go to iTunes University and you download the PDF files from the syllabus and the assignments and the video lectures are there, but it really felt like a dead course that you were resurrecting. And so, you know, you'd watch a few lectures and then work with it to fear and you'd say you'd stop. And then before you knew it, you know, you took a week off because you're busy and then you couldn't really remember what he said two weeks ago and then you don't really have anyone to ask to get a clarification on the assignment. So maybe you skipped that problem. And before you know it, you just say, this is just a waste of time and you just give up. It's not worth it. And what they did with these online education classes is they ran them like real classes. There's a bit of a gamification effect where they ran them for fixed intervals and they were deadlines and nothing motivates like a deadline. And you felt like people were watching and there was all this energy and activity in the forums and you can see there were thousands of people asking questions, everyone was synchronized. And at the end of it, the original Stanford classes, now they were offering you the certificate of completion, which they did, which by the way, wasn't on Stanford stationary because Stanford's okay, oh, we had nothing to do with this, you know? But even just these famous professors signing it was enough motivation and people were working for the certificate. They're working for the badge, you know, for the merit badge. And so I realized that they had cracked the code, right? And that people were excited about potentially completing the course and getting their little badge. And so we ran very similarly. And I'd say that we actually probably copied the techniques most closely of Udacity, which is the, because they have very good pedagogical techniques. They teach for the camera, they understand the teaching for online education is a different experience than teaching in a real classroom. You know, one of the problems a lot of these professors have is that they're famous professors. And so they say, yeah, I'll do an online class, this is what I'll do, I'll teach my class, put the camera in the back of the room and I hope you get what you get. And what really needs to be done is you need to view it like a one-on-one touring session in between you and the student and make it more personal. Yeah, it's kind of like taking your legacy app and making it mobile. It doesn't work. Exactly, you have to think about what's the medium. What's the medium? So bringing that pressure actually motivated those people. Yes, bringing the pressure motivates people. Yeah, people love the excitement and the energy. It's like a little Broadway performance when the class starts and then it kind of goes through. And so we did the same thing. And so we offer these certificates of completion. Now the relationship with Harvard and MIT is you're using the edX software. That's right. Is there any other collaboration going on there? We contribute back our improvements to the software so that they can use them with other partner universities. We, I think they like this a lot because they're open source and they're a nonprofit. And they loved what we were doing with open source. They love the way we live out in the open. They love the transparency that we allow ourselves in developing the software that developers can go in. They can see the bug reports. They can look at the source code. And they want the same type of thing for their educational stack. They don't want to be a close proprietary company. And so they were very excited to work with us because we know how to live on that edge between public and private where you're living out in the open but yet you might have to operate obviously coherently as a company. Do you see over time distributing the curriculum through other universities? Is that an opportunity for even globally? We'd certainly be open to it. Yeah, we'd be open to it to syndicating out the classes. So far we haven't had to because people seem to be coming to us. But our goal is to educate people about MongoDB on that side of it and just drive the adoption. So we have no problems with that, sure. So kind of playing off that your last answer there. So how do you align or how are the education efforts aligned with the larger goals of the company of growing the company and building a successful organization? Your CEO, Max Geerson, this morning mentioned that it's all free training education. So you're not monetizing the education which I think is a great thing. But how do you align the strategy? How do you align education with your larger core strategy? That's pretty straightforward. I mean, if you think about us as an open source company we have to create value and then to be a company that survives we have to harvest some value to make money. And the creation of the value is through the open source software and having a community addition. And education is part of the creation of that value. So by educating people how to use it and by reducing the friction to adopting the software we get more users. And by having a bigger community of users there's more potential people who might be interested in our subscription additions that are interested in proprietary features in those additions. So it just increases and enhances the size of the entire ecosystem from MongoDB. And I mean, this conference right today is in a sense an educational resource, right? For a lot of the attendees. That's right, that's right. And what you realize with tech companies is education is marketing in a lot of ways. And you know, these conferences they're very reasonably priced. They're made to be convenient. They're one day long. So they don't interrupt people's schedules. They're in a lot of different cities. And it's the same exact idea. And so I really think this education thing and the online education and the MOOCs are a natural successor and complement to what we're doing with community. And where do you see this going in terms of innovation of new ways to reach people? I mean, I'm sure you've got lots of ideas to expand this to even in new directions. What are some of the innovations you're looking to maybe implement and take this in another step? Yeah, I mean, MOOCs have taken the world by storm in the United States and there's a lot of excitement about them. I actually think that they're more interesting for professional development than they are for higher ed. And the reason is, if you look at any disruptive innovation, like you look at recorded music, 200 years ago, 150 years ago, all the music you were listening to, we were seeing live. And then when they came out with recorded music, which wasn't necessarily quite as good as live music, they didn't put it in front of audiences in Carnegie Hall. They would be pretty disappointed, right? They put it in places where normally music wasn't a possibility for you, like your living room and eventually your car, right? And I think that online education, it's the same thing. If you look at our online classes today, two thirds of the students are from South America and from Eastern Europe. And that's because these are the non-consumers. You know, these are the people who have no other venue or option. If your parents, if you can get into Harvard and your parents can pay for it, you should go. You know? I mean, it's great, right? Why not? But on the other hand, if you're in the workplace and you don't have time to go back for regular classes or go back to school, that's when this format works so well because it can fit in all the interstitial spaces of your life and you can advance your career that way. So is that typically the student you're trying to reach is really somebody who's in the midst of their career. They're not going to take two years off to go back to school. They're looking to develop their skills. They want to stay current. They know Mongo's important part of the growing and a big data management ecosystem. And so this is where that kind of can fit into their life in a way that makes sense to them and really provides some value. That's exactly right. I mean, and you know, we didn't know who would show up but that's who showed up. I'm excited about it because a lot of the places that people are taking the courses, we hardly even have business operations. Well, you know, we're driving adoption, getting that market ready for a day when we may come in and actually talk to them about what we have to offer on a premium side. Well, that struck me when you were talking. I mean, this is a great source of data for you about where the interest in Mongo is, how it's ebbing and flowing in different maybe geographic locations, maybe by vertical industries, whatever the case might be. And another way to use that data where it's a win-win for the student and a win for Mongo in the sense that you're able to mine that data and develop some services around that and find new markets to target. But that's exactly right. So the education piece is going well and I was super excited that it was so successful. I mean, I was very passionate about it coming in and of course you never really know, you know? And as an engineer, you're always a little bit field of dreams, right? You're like, you build it and they will come. And sometimes they do and sometimes they don't. Engineers and VCs. Exactly right, exactly right. And so about a month and a half ago, I took on this new role of GM of cloud services, which is something that we're just really dabbling in at this point. We've had a monitoring service that people don't know a lot about, but it's a free monitoring service for MongoDB. And we recently introduced a backup service as well, where we will back up your MongoDB databases, give you point-in-time restore, very reasonably priced, and you can turn it on in probably 25 minutes. Not really aimed at the enterprise, but instead aimed at businesses, start-ups and whatnot, that, you know, backup's not their core competency, it's not really, and they just want it to work. Right? Right. And I have a lot of, my background is in cloud services, so I was excited to get into that, you know, get back into that, that world of offering, you know, cloud services to users. So let's talk about that a little bit, because that's the right side of your brain. Yeah, that's right. With cloud services. So talk about what you guys are doing in cloud services. You got some relatively new services around backup and you've enhanced your monitoring, but talk about your role there and what TenGen does there. Yeah, so, you know, we're really building out the different tools that you need to run the database in the enterprise. So the Community Editions is a great database, it is missing certain features that enterprises tend to want around security and authentication. It's also missing, it was also, the Community Edition also doesn't have monitoring. So we went and we built out a monitoring component. We offer it in two different ways. If you're an enterprise customer, you can get the bits and you can run it in your own data center and if you're a smaller company, we'll monitor your database for you, for free. And then, which if you become a support customer, if you get a subscription and it comes in support, is great because it helps to support people to help you when you have a problem. And so a natural extension of that was this backup piece and to solve another part of the enterprise problem, which is backing up your database. Now, you know, backup is something that, you know, it's a little bit mundane and boring. It's like, oh, backup, who cares about backup? But it's also hard to do it well. Right? Everybody cares when you lose data. Yeah, everyone cares when you lose data. Exactly. First question you get. Are you backed up? Are you backed up? Ah! And you know, what people don't realize about backups is that most of the restores are not because of catastrophic failure. Yeah, it's humans. Most of the restores are because of human error, fat fingering as they say, messing something up, doing a bad release, and deleting data. And maybe that deletion is propagated amongst all your replica sets, and all of a sudden you've deleted it everywhere. And so, you know, I like to think of this, you know, window of vulnerability, which is how far behind are the backups? And a typical customer that does their own backups will simply take a snapshot every six hours. And that's fine, but the snapshot can often, there's also a large load of the system during that interval. And so let's, we said, well, if you really wanted to do it right, how would you do it right? And to do it right, you would have a continuous, you know, real-time backup. And so what our backup system does is it reads your replication log, and so we might be just minutes behind you. So if you actually, for some reason, lost everything, we could restore you to three minutes ago. So window of vulnerability, I like that term. A lot of people talk in terms of RPO, recovery point objective. Right. And they say that's the business term. And I go, what business person, you've ever met, talks about RPO, nobody. But window of vulnerability. It's a more intuitive term for me. That's an intuitive business term. Basically, it's like, how much data are you willing to lose? Exactly. Oh, I get it, okay. So you're basically saying you're taking people's window of vulnerability down from six hours, because maybe they back up twice a day, three times a day, four times a day, down to minutes, 15 minutes, less than that? It could be less than that. It could be less than that. It could be just a few minutes, you know, because we're real time taking their Oplog data. And the other thing about backup is, you never really know whether they're working. You know, like you're not confident that the backup's even working properly. I mean, you know, so you try to do it restore every once in a while. It's the kind of thing, you know, it's kind of like watching the instruments, you know, on an airplane. It's like nothing's ever happening. And so humans are particularly bad at this type of activity. You know, where the default is, nothing's going on. Well, I think that you're right on there, because one practitioner said to me, you know, all this data growth, how do you back up a petabyte? And he said to me, you don't, you know. He's right. You basically do, you know, Apple time machine for the enterprise. Right, no, you know, when you can't imagine taking a full L zero backup, when a data set gets large enough, and the way our backup system works is we take an initial sync. You got to seed it. We have to seed it. And after that, we never take a full sync again from them. We just take the incremental data from their Oplog and bring it into our data center. So how do you seed it? You have some kind of throttle or do you actually ship a disk? Well, it's an interesting question. We don't take ship disks today, but someday we may need to for big enough databases. Today, it just takes a while. Today, the initial sync is a full copy of their database over the internet. Okay. And then, you know, after that, we're taking permanently incremental data. So it's very carbonite-like. I mean, you start, sort of trickles in over a week or so. It is similar to what they do. The difference is, though, this is really built and tuned for MongoDB. And not for a laptop. Yeah, and it understands MongoDB's interface. It reads MongoDB's Oplog. We don't back up generic files. We can't back up your laptop. And carbonite actually has no module to back up MongoDB. So in that sense, it's not particularly competitive, but in the sense it's back up. But the concept of basically seeding it once and then taking it incremental daily. That's true. Yeah, that's very true. It is a similar concept. Yeah, good. Yeah. And does it tie into disaster recovery at all? Absolutely. I mean, for good DR, you want to be far enough away that when you lose your primary data center that you're going to be able to come back. And not only are we a data center that's far away from you, but we also run in multiple data centers and keep them far enough away from each other that it's DR compatible, if you will. So, but the further away you go, speed of light problems, right? The greater your window of vulnerability. But the truth is you're 80 milliseconds from Beijing. So that's not a real issue. It is true. There's an expression in engineering. You can buy bandwidth, but for latency, talk to God. And that is true. That is true. But it's mostly a bandwidth issue at the end of the day. Because let's say that we're really far away from the person. We're 80 milliseconds away. So that's not the real issue. The real issue is that as you get further away from a person, unless you're edge cashed and you're thinking about where to position yourself, you may not, you just may not have enough bandwidth to take that. You're not talking about transactions here, right? I mean, generally, right? You're not talking about ATM transactions, right? Well, most of our ATM transactions in particular are probably the least likely thing to be on MongoDB. Oh, that's what I'm saying, right? Yeah, but only because, and I think actually we have some features that could be used for ATM transactions. Today, not on MongoDB. But traditionally, these have gone to relational databases. If you think about relational transactions actually, and particularly the ATM problem, and I thought about this, you think about the number of people there are in the US and how often they take out money, it's actually not that many. And it's actually not that big a problem. I think what MongoDB really shines is in all of this machine generated data and all this observational data that works a lot faster than humans could type, right? Like, you know, the entire library of Congress, you know, could probably fit in a few terabytes. I'm probably even less than a terabyte. War and peace is under a megabyte, I think, right? And so anything that's human created is not that large. But then you look at something like machine generated data, like clickstream traffic or photos and videos, machines generating data from machines, and that's when it becomes just overwhelming. Yeah, absolutely. We were at the GE event earlier this week talking about the industrial internet and all this industrial equipment that's now creating, you know, it's sensor technology on jet engines, on wind turbines, on MRI machines, and you name it. And in healthcare, I mean, patients themselves are wearing the Fitbit and we're creating data on our person. So I mean, it's machine data clearly is really, I think where this is going next in terms of the ability to store it, optimize it, and optimize processes around the creation of that data. I think you're absolutely right. I mean, if you look on your iPhone today and you look at where your storage is going, you'll probably find that about three to 5% of it is data that you created by hand and the rest of it is machine sample data. I mean, music is machine sample data, right? People play music and then machine sampling. So, yeah, so most of the data in the world is machine sample. So Andrew, you're a startup guy. How did you end up in this role and so passionate about education and obviously the cloud piece? Yeah, yeah, I mean, I am a startup guy and this is in fact my first real job I've ever had. Before this, I started a couple of companies. My experience with the founders goes way back. I sold my first company to DoubleClick in 2000 and Dwight Merriman was my boss at the time and I had been following what was going on with TenGen and MongoDB. My first company was actually a little database company. It's called FlashBase, database is in a flash. So yeah, so this was a kind of a returning to my roots and the education thing was also a passion because I did a photo sharing startup. I have a lot of high end digital SLR equipment at home. I've always loved that side of things and the technical aspects, the precision of all that stuff and online ed was an opportunity to spend a tremendous amount of money on studio equipment, copy stands and high end cameras. You're talking our language now. Yeah, I love that stuff, I love that stuff. We have this lighting kit, by the way, just so you know, I recognize it. Yeah, right, it's nice that it's not too hot. So do you still pay attention or you must to the capital markets? I mean, what are you seeing there? In New York, California, you know. I don't pay a tremendous amount of attention to the capital markets. My only capital market, I worked at BlackRock when I first came to New York when it was a 200 person company managing 180 billion dollars. So you're not in touch with the VC scene in New York? Yeah, you know, I keep up with it but you know, I'm an obsession kind of guy and so whatever, you know, when I took some time off for the double click and I ran marathons and that was all I did. You know, people would see me at parties and they'd try to get anything out of me and all I could tell them was like the races I had run. And you know. That's great. And so I, you know, I don't obsess over that. I mean, I read the usual blogs, I probably read in the same blogs you do but you know, I'm not into that scene right now. Okay, good. All right, Andrew, listen, thanks very much for coming on theCUBE. Man of great passion and good luck with the journey and please come back to theCUBE. Love to have you again. All right, well, thank you very much. All right, pleasure. All right, everybody, keep it right there. Jeff Kelly and I will be back to wrap up the day. We're live in New York. This is theCUBE.