I had a chance to sit down for an exclusive conversation with the Dev Ittycheria, chief executive of MongoDB Inc., ahead of Amazon Web Services’ re:Invent conference, where SiliconANGLE and its video studio theCUBE will be providing wall-to-cloud coverage of the event.
We talked about how much AWS is changing and the role the partners ecosystem taking in the next generation cloud. We go into detail on the role of the changing independent software vendor and how AWS is enabling the innovation for its partners such as MongoDB.
The spotlight on partners and AWS co-existing together is a big theme this year at re:Invent. MongoDB’s experience and success with developers continues to this next-gen cloud evolution, in which there isn’t a one-size-fits-all data solution for customers. Ittycheria brings an insider perspective to the conversation this year at re:Invent.
The interview has been lightly edited for clarity.
Q: You guys are a top partner of AWS. I would imagine that you probably produce a lot of revenue for AWS because you really can’t turn off a database. How long has MongoDB been a partner with AWS? What’s the relationship?
A: The relationship is really strong. Actually, AWS spoke at one of our first user conferences in 2013. And since then, we’ve been working together. We’ve been at re:Invent since 2012 and have been a premier partner, an Emerald sponsor, for the last, four-to-five years. So we’re very committed to the relationship. We have a lot of things in common. We care a lot about customers. For us, our customers are the builders. We care a lot about removing friction from developers’ day-to-day work to be able to move fast and seize new opportunities or respond to new threats. Consequently, the partnership, obviously by nature of our common objectives, has really come together.
Q: How about MongoDB’s cloud journey? You look back at the history and you go back to the old LAMP stack days — the developer traction is really second to none. I remember the conversations, Dev, “MongoDB doesn’t scale.” I mean, every year we heard something along those lines but it turned out it was quite the opposite – it just kept scaling. I heard the same thing with AWS back in 2013, that “it’s really not for primetime.” It’s for hobbyists, not so much builders, maybe for startups, but that developer traction has translated. Can you take us through the journey of MongoDB — where it is now and kind of look back and tell us what’s state-of-the-art now?
A: Right — for those in your audience who don’t know too much about MongoDB, I’ll just start with the background. The company was founded by developers. It was basically the CTO and key developers from DoubleClick, who saw the challenges and limitations of the relational database architecture because they were trying to serve billions of ads per day and had to constantly work around the constraints of the relational database. So they decided, why don’t we just build a database that we’d want to use, and that was a catalyst to starting MongoDB.
The first thing they focused on was rather than having a tabular data structure, they chose a document data structure. Why documents? Because it’s much more natural and intuitive to work with data in documents in terms of how you can set parent-child relationships and how you just think about the relationship, the data is modeled in a more natural way in document than trying to connect data in hundreds of different tables. That enabled developers to just move so much faster.
The second thing that was focused on was building a truly distributed architecture, not some kind of add-on solution that made the existing architecture a little bit more scalable, but they really built, from the ground up, a truly distributed architecture. This allows you to do native replication, you can do sharding, and you can do it on a globally distributed basis. The document model is truly a superset of other models.
So since then, we’ve enabled other capabilities like search, you can still do joins, so you can do very transactionally intensive use cases with MongoDB. We’re fully ACID-compliant, so you have the highest forms of data guarantees. You can do very sophisticated things like time-series, you can do device synchronization. You can do real-time analytics because we can carve off read-only nodes to be able to read and query data in real-time rather than having to offload that data into a data warehouse. That enables developers to build a wide variety of applications on MongoDB and they get one unified and developer interface that’s highly elegant and seamless so the cost and tax of managing multiple single purpose tools goes away.
Q: You know, cloud adoption really is putting a lot of pressure on these systems, and we’re seeing companies in the ecosystem at AWS stepping up. You guys are doing a great job. We’re seeing a lot more acceleration around IT staying on-premises for certain use cases, yet you got the cloud as well growing for workloads and you get this hybrid steady state as an operational mode. I call that the classic cloud adoption track record. MongoDB is an example of multiple iterations in the cloud where you’re doing a lot more. We’re starting to see this tipping point with other partners and customers coming down that same pattern, building platforms on top of AWS, on top of the “primitives” as Adam Selipsky calls AWS’ foundational services. More horsepower, higher level services, industry specific capabilities with data are coming from all sides. This is a new kind of cloud, kind of an AWS next gen. You got the classic, high-performance infrastructure, it’s getting better and better. But now you’ve got this new application platform, that reminds me of the old ASP days if you will. Are you seeing customers doing things differently? Can you share your reaction to this new kind of SaaS platform that just isn’t an application, it’s more, it’s deeper than that? We call it supercloud at TheCube but what are you seeing?
A: We have over 37,000 customers of all shapes and sizes — from the largest companies in the world like Goldman Sachs and Intuit to cutting edge startups — who are building applications on MongoDB. Why did they choose MongoDB? Because it’s the fastest way to innovate. Developers find working with data so much easier using the document model than working on other types of architectures.
The document model is profoundly a breakthrough way to work with data and make it very, very easy for developers to move fast to build these modern applications; that is, applications architected with microservices, event-driven architectures, and addressing sophisticated use cases like time-series, etc. And then ultimately, now they’re building smarter apps by embedding machine learning. We have a bunch of companies building machine learning applications on top of MongoDB, such as Rent the Runway, Bosch, and Mount Sinai Health System.
The reason they’re doing that is they get the benefits of being able to build and work with data faster than any other platform, and it’s highly scalable and performant in a way that no other platform offers. They can run their workloads both locally in one availability zone or they could be anywhere in the world, over 100 different regions across multiple clouds.
Q: Let’s start with the performance side. AWS is really working on physics — they want to squeeze as much energy out of their chips. I’ve never met a developer that said they want to run their workload on a slower platform or on slower hardware. You guys have a lot of experience tuning in with Graviton instances, we’re seeing a lot more AWS EC2 instances. We’re seeing a lot more integrated end-to-end stories, data is now security. A lot is going on around the hardware performance specialization and we’re really seeing a modern data stack emerging. What are your thoughts on that?
A: If you had asked me when the cloud started going vogue in the later part of the last decade and told me, sitting here 12-15 years later, that we would be talking about chip processing speeds, I would’ve thought, “Nah we would’ve moved on by then.”
What’s really clear is that, to your point, customers care about performance, they care about price performance. AWS has significant investments in building their own chips, like Graviton, for improving price performance and efficiency. We have actually deployed a significant portion of our Atlas fleet on AWS Graviton-based instances. They’ve built other chipsets for AI, ML, and scientific applications, like Trainium for training models and Inferentia for running inferences. They’re doing things like Nitro, to improve security and manageability of EC2 instances.
What that really speaks to is that the cloud providers, like AWS, are focusing on the price performance of their primitives and their infrastructure layer. If you look at their revenue, probably about 60% to 70% of revenue comes from pure infrastructure. To your point, they can’t offer a second-class infrastructure solution and still win. Given that, now they’re seeing a lot of competition from Azure which is building its own chipsets as well, and Google’s already obviously doing that by building specialized chipsets for machine learning. You’re seeing these cloud providers compete to make their platform the most performant and the most price competitive in the marketplace. Which in turn, allows us a great technology base to build on to enable developers to build these highly performant applications that customers now demand.
Q: I think that’s a really great point. It’s so funny, Dev, because we don’t talk speeds and feeds anymore. We’re not talking about boxes. That’s all kind of old school thinking because it was a data center mentality — speeds and feeds — and that was super important. We’re kind of coming back to that in the cloud now with distributed architecture. As you put your platforms out there for developers, you have to run fast. You can’t give the developer subpar or any kind of performance system, they’ll go somewhere else. That’s the reality of developers.
So you’ve got the tale of two clouds going on here. You’ve got Amazon Classic — IaaS, keep making it faster under the hood, and then you got the new abstraction layers and the higher-level services. That’s where you guys are bridging this new generational shift. Where you can run a headless application, you can run a SaaS app as refactored with data. You see a lot more innovation with developers running stuff in the CI/CD pipeline that was once IT. And you’re seeing security and data operations kind of emerging as a structural change of how companies are transforming on the business side. What’s your reaction to that business transformation and the role of the developer?
A: I have to give kudos to AWS for being very customer obsessed and innovative, now with something like over 300 services. Obviously, they’re the ones who created the cloud industry and they continue to push innovation in the space. Every startup today is building with the cloud because they have so many building blocks to start with. But what we have found from talking to our customers is that in some ways, the onus is on the customer to figure out which building block to use to be able to stitch together the applications and solutions they want to build. What we have done is taken an opinionated point of view and said, “we will enable you to do that using one data platform.”
AWS today offers about a dozen different types of databases. Based on our conversations with customers, we don’t think that having a tool for every job makes sense because over time, the tax and cost of learning, managing and supporting those different tools doesn’t make a lot of sense or just becomes cost prohibitive. We believe offering one data platform, one elegant user experience, one way to address the broadest set of use cases is a better way. But clearly customers have a choice. They can use AWS’ primitives, and the second layer of services as you described, or they can use us. Fortunately, we’ve seen a lot of customers come to us with our approach.
Still, I have to give kudos to AWS for being very customer obsessed. We have a great relationship with them, both technically in terms of the product integrations we do, as well as working with them in the field, on joint-customer opportunities.
Q: How is the AWS Marketplace relationship going? Some of the partners are really seeing great economic and cross selling, joint selling, or AWS selling partner software. There’s a real revenue opportunity with that relationship. Can you comment on that?
A: We have been working as partners in the AWS Marketplace for many years now, more from a field point of view where customers can leverage their existing commitments to AWS and use Atlas which gets credited to their commits. There are also some sales incentives for people in the field to work together so that everyone wins should we collectively win a customer. What we’ve recently announced is this Pay-As-You-Go initiative where a customer on the AWS Marketplace can basically turn up an Atlas instance with no commitment upfront. So we’re continually pushing the envelope to reduce the friction for people to use Atlas on AWS and it’s working really, really well. The uptake has been very strong and we feel like we’re just getting started because we’re so excited about the results we’re seeing.
Q: One of the underlying keynote themes, I think, is developer productivity. You said making things easy is a big deal; self-service, getting in and trying —t hese are what developer-friendly tools are like. So I have to ask you, because this comes up a lot. If you take the digital transformation concept to its completion, as a thought exercise, you completely transform a company with technology. That’s the business transformation outcome. Take it to completion. What does that look like? If you go there, you’d say the company is the app. The company is the data. It’s not a department serving the business, it’s the business. This is what we’re seeing as the next big mountain to climb, which is companies that transform themselves are technology companies — not a department like IT. What’s your view on this?
A: I’ve had the good fortune to be able to talk to thousands of customers all over the world. And you know one thing, John, they never tell me that they’re innovating too quickly. In fact, they always tell me the reverse. They tell me all the obstacles and impediments they have to be able to move fast. One of the reasons they gravitate to MongoDB is the speed with which they can build applications, to drive greater developer productivity. By definition, developer productivity is a proxy for innovation. The faster you can help your developers move, the faster they can push out code. The faster they can iterate and build new solutions, or add more capabilities on existing applications. The faster you can innovate to seize new opportunities or respond to new threats in your business. That resonates with every C-level executive. To your point, developers are not some side hustle that they think about once a while, they’re core to the business. So developers have amassed an enormous amount of power and influence. Their engineering teams are front and center in terms of how they think about building capabilities and building their business. That’s also enabled now that every company is becoming a software company, because it all starts with software. Software enables, defines or creates almost every company’s value proposition.
Q: It makes me smile because I love operating systems. One of my hobbies in college was systems programming. Now we’re kind of like the operating systems in the cloud — everything’s got specialized capabilities. That’s a big theme here at re:Invent. If you look at the announcements that traditionally come Monday night with Peter DeSantis, with new instances, new chips, so this whole specialized component is like an engine — you’ve got a core and other subsystems. This is going to be an integral part of how companies architect their platform or, as Adam [Selipsky] calls it, “the landing zone.” You’re going to start seeing a new architectural thinking for companies. Can you share your experience on how companies should look at this opportunity as a plethora of more goodness on the hardware? Because now you can mix and match, everything you need to build foundational, high performance capabilities.
A: This is where AWS is really enabling all companies, including companies like MongoDB, to push the envelope in innovation. We’ve seen two big platform shifts over the last 15 years with mobile and the shift to cloud. I believe the next big platform shift is moving from dumb apps to smart apps that incorporate machine learning, AI and very sophisticated automation. When you start automating human decision making, rather than looking at a dashboard and saying, “Okay, I see the data now, now I have to do this”, you can automate that into your applications, leveraging real-time data analytics, and they become that much smarter. Ultimately, that becomes a developer challenge that MongoDB can solve. We feel really good about our position and taking advantage of those next big trends in software such as leveraging the price performance curves that AWS continues to push in terms of their hardware performance, networking performance, price performance of storage to build those next generation of modern applications.
Q: OK, so let me get this straight. You have next generation intelligent smart apps, and you have AI-generative solutions coming out around the corner. This is a pretty good position for MongoDB to be in with data. I mean, this is what you do. You’re in that state of the action.The world is starting to wake up to this. What’s it like?
A: We’re really excited and bullish about the future. We think we’re well positioned because we have amassed an amazing amount of developer mindshare. We are the most popular modern data platform in the world, and there are developers in almost every corner of the planet using us to do something spectacular. We think the more AI becomes democratized – accessible by all developers vs. only the most accomplished data scientists – by essentially delivering the right tools that enable developers to build these very sophisticated, smart applications, we are well positioned. That’s obviously going to be a focus for us. I think this is going to be a 10- to 15-year run and we’re just getting started in this whole area.
Q: I think you guys are really well-positioned and that’s a great point. Adam [Selipsky, AWS CEO] mentioned to me in my interview that the role of a data analyst kind of goes away. Everyone’s a data analyst, or we’ll see a specialization in core data engineering, like an SRE role, but Data Ops and data-as- code is a big deal. Adam is a great leader. He’s going to help educate customers to use technology for business advantage and transformation. Andy [Jassy] did a great job of making technology great and innovative and setting the table. Adam’s got to continue to bring it to enterprises and businesses. It’s gonna be an interesting point in time we’re in now. How would you categorize this point in time if you had a bumper sticker?
A: The tech world is pivoting towards what I call rationalization or cost optimization. Over the last 10 years it’s been all about speed, speed, speed to move as fast as possible. People still value speed, but they want to do it at some sort of predictable cost model. You’re going to see a lot more focus around cost and cost optimization. That’s where we think having one platform, by definition, enables vendor consolidation in way for people to cut costs so that they can still move fast, but they don’t have to incur the tax of using a whole bunch of different point tools. So we’re well-positioned. The bumper stickers for re:Invent 2022, and I might reference two options, are “Build more with less [cost/complexity]” or “Build the next big thing with MongoDB.”