Atlas 5-Year Anniversary Podcast Series Episode 1 - Onramp to Atlas
Rate this podcast
In this series, my co-hosts, Jesse Hall, and Nic Raboy will talk with some of the people responsible for building, and launching the platform that helped to transform MongoDB as a company.
In episode 2, Zero to Database as a Service, we’ll chat with Cailin Nelson, SVP of Engineering, and Cory Mintz, VP of Engineering - about Atlas as a product and how it was built and launched.
In episode 3, we’ll Go Mobile, talking with Alexander Stigsen, Founder of the Realm Mobile Database which has become a part of the Atlas Platform.
In episode 4, we’ll wrap the series up with a panel discussion and review some of our valued customer comments about the platform.
Without further adue, here is the transcript from episode one of this series.
Sahir: [00:00:00] Hi Everyone. My name is Sahir Azam and I'm the chief product officer at Mongo DB. Welcome to the Mongo DB podcast.
Mike: [00:00:07] Okay. Today, we're going to be talking about Mongo to be Atlas and the journey that has taken place to bring us to this point, the five-year anniversary of MongoDB Atlas of a launch of MongoDB Atlas. And I'm joined in the studio today by a couple of guests. And we'll start by introducing Sahir Azam chief product officer at Mongo DB. Sahir, welcome to the show. It's great to have you on the podcast.
Sahir: [00:00:31] Hey, Hey Mike. Great to be here.
Mike: [00:00:33] Terrific. And we're also joined by Andrew Davidson. Andrew is vice-president of product cloud products at Mongo DB. Is it, do I have that right?
Andrew: [00:00:41] That's right? Good to be here, Mike. How you doin?
Mike: [00:00:44] Doing great. It's great to have you on the show. And of course are my co-hosts for the day. Is Jesse Hall also known as codeSTACKr. Welcome back to the show, Jesse. It's great to have you on
Jesse: [00:00:54] I'm fairly new here. So I'm excited to hear about the, history of Atlas
Mike: [00:00:58] fantastic. Yeah. W we're gonna, we're gonna get into that. But before we do so here, I guess we'll maybe introduce yourself to the audience, talk a little bit about who you are and what you.
Sahir: [00:01:09] Yeah. So, I mentioned earlier, I run the product organization at Mongo and as part of my core focus, I think about the products we build the roadmaps of those products and how they serve customers and ultimately help us grow our business. And I've been with the company for about five years. Coincidentally I was recruited to lead basically the transition of the business. Open source enterprise software company to becoming a SAS vendor. And so I came on right before the launch of Atlas, Andrew on the line here certainly has the history of how Atlas came to be even prior to me joining. But, uh, it's been a heck of a ride.
Mike: [00:01:46] Fantastic. Well, Andrew, that brings us to you once, yet, let folks know who you are and what you do.
Andrew: [00:01:52] sure. Yeah. Similar to Sahir, I focus on product management, but a really more specifically focused on our cloud product suite. And if you think about it, that was something that five years ago, when we first launched Atlas was just an early kernel, a little bit of a startup inside of our broader company. And so before that time, I was very focused on our traditional more private cloud management experience from marketing the and it's really just been this amazing journey to really transform this company with Sahir and so many others into being a cloud company. So really excited to be here on this milestone.
Mike: [00:02:25] Fantastic. And Jesse, so you've been with Mongo to be, I guess, relatively the least amount of time among the four of us, but maybe talk about your experience with Mongo to be and cloud in general.
Jesse: [00:02:36] Yeah. So I've used it several times in some tutorials that I've created on the Atlas portion of it. Going through the onboarding experience and learning how it actually works, how the command line and all of that was amazing to understand it from that perspective as well. So, yeah, I'm excited to see how you took it from that to the cloud.
Mike: [00:02:58] Yeah. Yeah, me too. And if you think about the journey I'm going to be was a successful open source product. I was a project that was largely used by developers. To increase agility. It represented a different way to store data and it wasn't a perfect journey. There were some challenges early on, specifically around the uniqueness of the mechanism that it's using to store data is different from traditional forms. And. So I guess Andrew you've been here the longest over eight years. Talk about the challenges of transitioning from a software product to an online database, as a service.
Andrew: [00:03:37] Yeah. Sure. When you think back to where we were, say eight and a half years ago, to your point, we had this kind of almost new category of data experience for developers that gave them this natural way to interface with data in a way that was totally reflective of the way they wanted to think about their data, the objects in there. And we came in and revolutionized the world with this way of interfacing with data. And that's what led to them. I'm going to be just exploding in popularity. It was just mind boggling to see millions of people every month, experiencing MongoDB for the first time as pure open source software on their laptops. But as we move forward over the years, we realized. We could be this phenomenal database that gave developers exactly the way they want to interface with data. We could be incredibly scalable. We could go up to any level of scale with vertical and horizontal kind of linear cost economics, really built for cloud. We could do all of that, but if our customers continued to have to self manage all of this software at scale, we realized, frankly, we might get left behind in the end. We might get beaten by databases that weren't as good. But then we're going to be delivered at a higher level of abstraction, fully managed service. So we went all in as a company recognizing we need to make this just so easy for people to get started and to go up to any level of scale. And that's really what Atlas was about. It was all about democratizing this incredible database, which had already democratize a new data model, but making it accessible for production use cases in the cloud, anywhere in the room. And I think when you see what's happened today with just millions of people who have now used Atlas, the same magnitude of number of have had used our self-managed software. It's just amazing to see how far.
Mike: [00:05:21] Yeah. Yeah. It's been quite a ride and it is interesting timing. So here, so you joined right around the same time. I think it was, I think a couple of months prior to the launch of Atlas. Tell us about like your role early.
Sahir: [00:05:36] Yeah, I think what attracted me to Mongo DB in the first place, certainly the team, I knew there was a strong team here and I absolutely knew of the sort of popularity and. Just disruption that the open source technology and database had created in the market just as, somebody being an it and technology. And certainly it'd be hard to miss. So I had a very kind of positive impression overall of the business, but the thing that really did it for me was the fact that the company was embarking on this strategic expansion to become a SAS company and deliver this database as a service with Atlas, because I had certainly built. In my own mind sort of conviction that for open source companies, the right business model that would ultimately be most successful was distributing tech technology as a matter of service so that it can get the reach global audiences and, really democratize that experiences as Andrew mentioned. So that was the most interesting challenge. And when I joined the company, I think. Part of everyone understands is okay, it's a managed version of bongo DB, and there's a whole bunch of automation, elasticity and pay as you go pricing and all of the things that you would expect in the early days from a managed service. But the more interesting thing that I think is sometimes hidden away is how much it's really transformed Mongo DB. The company's go to market strategy. As well, it's allowed us to really reach, tens of thousands of customers and millions of developers worldwide. And that's a function of the fact that it's just so easy to get started. You can start off on our free tier or as you start building your application and it scales just get going on a credit card and then ultimately engaged and, in a larger level with our organization, as you start to get to mission criticality and scale. That's really hard to do in a, a traditional sort of enterprise software model. It's easy to do for large customers. It's not easy to do for the broad base of them. Mid-market and the SMB and the startups and the ecosystem. And together with the team, we put a lot of focus into thinking about how do we make sure we widen the funnel as much as possible and get as many developers to try Atlas as the default experience we're using Mongo DB, because we felt a, it was definitely the best way to use the technology, but also for us as a company, it was the most powerful way for us to scale our overall operations.
Mike: [00:07:58] Okay.
Jesse: [00:08:00] Okay.
Mike: [00:08:00] So obviously there's going to be some challenges early on in the minds of the early adopters. Now we've had some relatively large names. I don't know if we can say any names of customers that were early adopters, but there were obviously challenges around that. What are some of the challenges that were particularly difficult when you started to talk to some of these larger name companies? What are some of the things that. Really concerned about early on.
Sahir: [00:08:28] Yeah I'll try them a little bit. And Andrew, I'm sure you have thoughts on this as well. So I think in the, when we phased out sort of the strategy for Atlas in the early years, when we first launched, it's funny to think back. We were only on AWS and I think we were in maybe four or five regions at the time if I remember correctly and the first kind of six to 12 months was really optimized for. Let's call it lower end use cases where you could come in. You didn't necessarily have high-end requirements around security or compliance guarantees. And so I think the biggest barrier to entry for larger customers or more mission critical sort of sensitive applications was. We as ourselves had not yet gotten our own third-party compliance certifications, there were certain enterprise level security capabilities like encryption, bring your own key encryption things like, private networking with with peering on the cloud providers that we just hadn't built yet on our roadmap. And we wanted to make sure we prioritize correctly. So I think that was the. Internal factor. The external factor was, five years ago. It wasn't so obvious that for the large enterprise, that databases of service would be the default way to consume databases in the cloud. Certainly there was some of that traction happening, but if you look at it compared to today, it was still early days. And I laugh because early on, we probably got positively surprised by some early conservative enterprise names. Maybe Thermo Fisher was one of them. We had I want to say AstraZeneca, perhaps a couple of like really established brand names who are, bullish on the cloud, believed in Mongo DB as a key enabling technology. And in many ways where those early partners with us in the enterprise segment were to help develop the maturity we needed to scale over time.
Mike: [00:10:23] Yeah,
Andrew: [00:10:23] I remember the, these this kind of wake up call moment where you realized the pace of being a cloud company is just so much higher than what we had traditionally been before, where it was, a bit more of a slow moving enterprise type of sales motion, where you have a very big, POC phase and a bunch of kind of setup time and months of delivery. That whole model though, was changing. The whole idea of Atlas was to enable our customer to very rapidly and self-service that service matter build amazing applications. And so you had people come in the matter of hours, started to do really cool, amazing stuff. And sometimes we weren't even ready for that. We weren't even ready to be responsive enough for them. So we had to develop these new muscles. Be on the pulse of what this type of new speed of customer expected. I remember in one of our earliest large-scale customers who would just take us to the limits, it was, we had, I think actually funny enough, multiple cricket league, fantasy sports apps out of India, they were all like just booming and popularity during the India premier league.
Mike: [00:11:25] Okay.
Andrew: [00:11:26] Cricket competition. And it was just like so crazy how many people were storming into this application, the platform at the same time and realizing that we had a platform that could, actually scale to their needs was amazing, but it was also this constant realization that every new level of scale, every kind of new rung is going to require us to build out new operational chops, new muscles, new maturity, and we're still, it's an endless journey, a customer today. A thousand times bigger than what we could accommodate at that time. But I can imagine that the customers of, five years from now will be a, yet another couple of order magnitude, larger or orders meant to larger. And it's just going to keep challenging us. But now we're in this mindset of expecting that and always working to get that next level, which is exciting.
Mike: [00:12:09] Yeah. I'm sure it hasn't always been a smooth ride. I'm sure there were some hiccups along the way. And maybe even around scale, you mentioned, we got surprised. Do you want to talk a little bit about maybe some of that massive uptake. Did we have trouble offering this product as a service? Just based on the number of customers that we were able to sign up?
Sahir: [00:12:30] I'd say by and large, it's been a really smooth ride. I think one of the ones, the surprises that kind of I think is worth sharing is we have. I think just under or close to 80 regions now in Atlas and the promise of the cloud at least on paper is endless scale and availability of resources, whether that be compute or networking or storage. That's largely true for most customers in major regions where the cloud providers are. But if you're in a region that's not a primary region or you've got a massive rollout where you need a lot of compute capacity, a lot of network capacity it's not suddenly available for you on demand all the time. There are supply chain data center or, resources backing all of this and our partners, do a really great job, obviously staying ahead of that demand, but there are sometimes constraints. And so I think we reached a certain scale inflection point where we were consistently bumping up. The infrastructure cloud providers limits in terms of availability of capacity. And, we've worked with them on making sure our quotas were set properly and that we were treated in a special case, but there were definitely a couple of times where, we had a new application launching for a customer. It's not like it was a quota we were heading there literally was just not there weren't enough VMs and underlying physical infrastructure is set up and available in those data centers. And so we had some teething pains like working with our cloud provider friends to make sure that we were always projecting ahead with more and more I think, of a forward look to them so that we can, make sure we're not blocking our customers. Funny cloud learnings, I would say.
Mike: [00:14:18] Well, I guess that answers that, I was going to ask the question, why not? Build our own cloud, why not build, a massive data center and try and meet the demands with something like, an ops manager tool and make that a service offering. But I guess that really answers the question that the demand, the level of demand around the world would be so difficult. Was that ever a consideration though? Building our own
Sahir: [00:14:43] so ironically, we actually did run our own infrastructure in the early days for our cloud backup service. So we had spinning disks and physical devices, our own colo space, and frankly, we just outgrew it. I think there's two factors for us. One, the database is pretty. Low in the stack, so to speak. So it needs to, and as an operational transactional service, We need to be really close to where the application actually runs. And the power of what the hyperscale cloud providers has built is just immense reach. So now any small company can stand up a local site or a point of presence, so to speak in any part of the world, across those different regions that they have. And so the idea that. Having a single region that we perhaps had the economies of scale in just doesn't make sense. We're very dispersed because of all the different regions we support across the major cloud providers and the need to be close to where the application is. So just given the dynamic of running a database, the service, it is really important that we sit in those public major public cloud providers, right by the side, those those customers, the other. Is really just that we benefit from the innovation that the hyperscale cloud providers put out in the market themselves. Right. There's higher levels of abstraction. We don't want to be sitting there. We have limited resources like any company, would we rather spend the dollars on racking and stacking hardware and, managing our own data center footprint and networking stack and all of that, or would we rather spend those reasons? Consuming as a service and then building more value for our customers. So the same thing we, we just engage with customers and why they choose Atlas is very much true to us as we build our cloud platforms.
Andrew: [00:16:29] Yeah. I If you think about it, I'm going to be is really the only company that's giving developers this totally native data model. That's so easy to get started with at the prototyping phase. They can go up to any level of scale from there that can read and write across 80 regions across the big three cloud providers all over the world. And for us to not stay laser-focused on that level. Making developers able to build incredible global applications would just be to pull our focus away from really the most important thing for us, which is to be obsessed with that customer experience rather than the infrastructure building blocks in the backend, which of course we do optimize them in close partnership with our cloud provider partners to Sahir's point.. .
Jesse: [00:17:09] So along with all of these challenges to scale over time, there was also other competitors trying to do the same thing. So how does Mongo DB, continue to have a competitive advantage?
Sahir: [00:17:22] Yeah, I think it's a consistent investment in engineering, R and D and innovation, right? If you look at the capabilities we've released, the core of the database surrounding the database and Atlas, the new services that integrated simplify the architecture for applications, some of the newer things we have, like search or realm or what we're doing with analytics with that was data lake. I'll put our ability to push out more value and capability to customers against any competitor in the world. I think we've got a pretty strong track record there, but at a more kind of macro level. If you went back kind of five years ago to the launch of Atlas, most customers and developers, how to trade off to make you either go with a technology that's very deep on functionality and best of breed. So to speak in a particular domain. Like a Mongo DB, then you have to, that's typically all software, so you tend to have to operate it yourself, learn how to manage and scale and monitor and all those different things. Or you want to choose a managed service experience where you get, the ease of use of just getting started and scaling and having all the pay as you go kind of consumption models. But those databases are nowhere close to as capable as the best of breed players. That was the state of the mark. Five years ago, but now, fast forward to 2021 and going forward customers no longer have to make that trade. You have multicloud and sort of database and service offerings analytics as a service offerings, which you learning players that have not only the best of breed capability, that's a lot deeper than the first party services that are managed by the cloud providers, but are also delivered in this really amazing, scalable manner. Consumption-based model so that trade-off is no longer there. And I think that's a key part of what drives our success is the fact that, we have the best capabilities. That's the features and the costs that at the cost of developers and organizations want. We deliver it as a really fluid elastic managed service. And then guess what, for enterprises, especially multicloud is an increasingly strategic sort of characteristic they look for in their major providers, especially their data providers. And we're available on all three of the major public clouds with Atlas. That's a very unique proposition. No one else can offer that. And so that's the thing that really drives in this
Mike: [00:19:38] Yeah.
Sahir: [00:19:39] powering, the acceleration of the Atlas business.
Mike: [00:19:42] Yeah. And so, Andrew, I wonder if for the folks that are not familiar with Atlas, the architecture you want to just give an overview of how Atlas works and leverages the multiple cloud providers behind the scenes. Andrew: [00:19:56] Yeah, sure. Look, anyone who's not used not going to be Atlas, I encourage you just, sign up right away. It's the kind of thing where in just a matter of five minutes, you can deploy a free sandbox cluster and really start building your hello world. Experience your hello world application on top of MongoDB to be the way Atlas really works is essentially we try and make it as simple as possible. You sign up. Then you decide which cloud provider and which region in that cloud provider do I want to launch my database cluster into, and you can choose between those 80 regions to hear mentioned or you can do more advanced stuff, you can decide to go multi-region, you can decide to go even multicloud all within the same database cluster. And the key thing is that you can decide to start really tiny, even at the free level or at our dedicated cluster, starting at $60. Or you can go up to just massive scale sharded clusters that can power millions of concurrent users. And what's really exciting is you can transition those clusters between those states with no downtime. At any time you can start single region and small and scale up or scale to multiple regions or scale to multiple clouds and each step of the way you're meeting whatever your latest business objectives are or whatever the needs of your application are. But in general, you don't have to completely reinvent the wheel and rearchitect your app each step of the way. That's where MongoDB makes it just so as you to start at that prototyping level and then get up to the levels of scale. Now on the backend, Atlas does all of this with of course, huge amount of sophistication. There's dedicated virtual, private clouds per customer, per region for a dedicated clusters. You can connect into those clusters using VPC, Piering, or private link, offering a variety of secure ways to connect without having to deal with public IP access lists. You can also use the. We have a wide variety of authentication and authorization options, database auditing, like Sahir mentioned, bring your own key encryption and even client-side field level encryption, which allows you to encrypt data before it even goes into the database for the subsets of your schema at the highest classification level. So we make it, the whole philosophy here is to democratize making it easy to build applications in a privacy optimized way to really ultimately make it possible, to have millions of end consumers have a better experience. And use all this wonderful digital experiences that everyone's building out there.
Jesse: [00:22:09] So here we talked about how just the Mongo DB software, there was a steady growth, right. But once we went to the cloud with Atlas, the success of that, how did that impact our business?
Sahir: [00:22:20] Yeah, I think it's been obviously Quite impactful in terms of just driving the acceleration of growth and continued success of MongoDB. We were fortunate, five, six years ago when Atlas was being built and launched that, our business was quite healthy. We were about a year out from IPO. We had many enterprise customers that were choosing our commercial technology to power, their mission, critical applications. That continues through today. So the idea of launching outlets was although certainly strategic and, had we saw where the market was going. And we knew this would in many ways, be the flagship product for the company in the term, it was done out of sort of an offensive view to getting to market. And so if you look at it now, Atlas is about 51% of our revenue. It's, the fastest growing product in our portfolio, Atlas is no longer just a database. It's a whole data platform where we've collapsed a bunch of other capabilities in the architecture of an application. So it's much simpler for developers. And over time we expected that 51% number is only going to continue to be, a larger percentage of our business, but it's important to know. Making sure that we deliver a powerful open source database to the market, that we have an enterprise version of the software for customers who aren't for applications or customers that aren't yet in the crowd, or may never go to the cloud for certain workloads is super critical. This sort of idea of run anywhere. And the reason why is, oftentimes. Timeline for modernizing an application. Let's say you're a large insurance provider or a bank or something. You've got thousands of these applications on legacy databases. There's an intense need to monitor modernize. Those that save costs to unlock developer, agility, that timeline of choosing a database. First of all, it's a decision that lasts typically seven to 10 years. So it's a long-term investment decision, but it's not always timed with a cloud model. So the idea that if you're on premises, that you can modernize to an amazing database, like Mongo DB, perhaps run it in Kubernetes, run it in virtual machines in your own data center. But then, two years later, if that application needs to move to the cloud, it's just a seamless migration into Atlas on any cloud provider you choose. That's a very unique and powerful, compelling story for, especially for large organizations, because what they don't want is to modernize or rewrite an application twice, once to get the value on pro-business and then have to think about it again later, if the app moves to the cloud, it's one seamless journey and that hybrid model. Of moving customers to words outlets over time is really been a cohesive strategies. It's not just Atlas, it's open source and the enterprise version all seamlessly playing in a uniform model.
Mike: [00:25:04] Hmm. Fantastic. And, I love that, the journey that. Atlas has been on it's really become a platform. It's no longer just a database as a service. It's really, an indispensable tool that developers can use to increase agility. And, I'm just looking back at the kind of steady drum beat of additional features that have been added to, to really transform Atlas into a platform starting with free tier and increasing the regions and the coverage and. Client side field level encryption. And just the list of features that have been added is pretty incredible. I think I would be remiss if I didn't ask both of you to maybe talk a little bit about the future. Obviously there's things like, I don't know, like invisibility of the service and AI and ML and what are some of the things that you're thinking about, I guess, without, tipping your cards too much. Talk about what's interesting to you in the future of cloud.
Andrew: [00:25:56] I'll take a quick pass. Just I love the question to me, the most important thing for us to be just laser focused on always going forward. Is to deliver a truly integrated, elegant experience for our end customers that is just differentiated from essentially a user experience perspective from everything else that's out there. And the platform is such a fundamental part of that, being a possibility, it starts with that document data model, which is this super set data model that can express within it, everything from key value to, essentially relational and object and. And then behind making it possible to access all of those different data models through a single developer native interface, but then making it possible to drive different physical workloads on the backend of that. And what by workloads, I mean, different ways of storing the data in different algorithms used to analyze that data, making it possible to do everything from operational transactional to those search use cases to here mentioned a data lake and mobile synchronization. Streaming, et cetera, making all of that easily accessible through that single elegant interface. That is something that requires just constant focus on not adding new knobs, not adding new complex service area, not adding a millions of new permutations, but making it elegant and accessible to do all of these wonderful data models and workload types and expanding out from there. So you'll just see us keep, I think focusing. Yeah.
Mike: [00:27:15] Fantastic. I'll just give a plug. This is the first in the series that we're calling on ramp to Mongo to be Atlas. We're going to go a little bit deeper into the architecture. We're going to talk with some engineering folks. Then we're going to go into the mobile space and talk with Alexander Stevenson and and talk a little bit about the realm story. And then we're going to wrap it up with a panel discussion where we'll actually have some customer comments and and we'll provide a little bit. Detail into what the future might look like in that round table discussion with all of the guests. I just want to thank both of you for taking the time to chat with us and I'll give you a space to, to mention anything else you'd like to talk about before we wrap the episode up. Sahir, anything?
Sahir: [00:27:54] Nothing really to add other than just a thank you. And it's been humbling to think about the fact that this product is growing so fast in five years, and it feels like we're just getting started. I would encourage everyone to keep an eye out for our annual user conference next month. And some of the exciting announcements we have and Atlas and across the portfolio going forward, certainly not letting off the gas.
Mike: [00:28:15] Great. Any final words Andrew?
Andrew: [00:28:18] yeah, I'll just say, mom going to be very much a big ten community. Over a hundred thousand people are signing up for Atlas every month. We invest so much in making it easy to absorb, learn, dive into to university courses, dive into our wonderful documentation and build amazing things on us. We're here to help and we look forward to seeing you on the platform.
Mike: [00:28:36] Fantastic. Jesse, any final words?
Jesse: [00:28:38] No. I want just want to thank both of you for joining us. It's been very great to hear about how it got started and look forward to the next episodes.
Mike: [00:28:49] right.
Sahir: [00:28:49] Thanks guys.
Mike: [00:28:50] Thank you.
How to Use Custom Archival Rules and Partitioning on MongoDB Atlas Online Archive
Jun 07, 2022