Does MongoDB regret offending the open-source community? Not one bit.
Surging sales of MongoDB Atlas show the database upstart is more than holding its own against cloud competitors.
Two years ago, MongoDB CEO Dev Ittycheria put himself in the middle of a raging debate about what it means to be an open-source company in the cloud computing era. It's fair to say he hasn't looked back.
Earlier this month, the New York-based database company reported overall revenue growth of 46%, beating analyst expectations on the back of 76% revenue growth from its flagship product, the managed MongoDB Atlas database. The company's stock is up 776% since its 2017 IPO, and up 203% since it announced in October 2018 that it would be changing the terms of the license that governs the free open-source MongoDB project.
That change, which followed a similar change introduced by Redis Labs, another database company, sparked controversy over whether open-source projects that place restrictions on how they can be used deserve the backing of the open-source community — a community from which MongoDB was born, and one that forever changed enterprise software over the last decade. The change was designed to hinder cloud providers such as AWS from offering their own managed versions of the MongoDB open-source project as a commercial service, which is what MongoDB does with its Atlas product.
A longstanding complaint raised by Ittycheria, and executives at other companies built around open-source projects, is that AWS and other cloud providers unfairly took advantage of the work they did in building those open-source projects when offering their own revenue-generating managed services. (Though the practice is perfectly legal under the permissive licensing terms employed by most of those open-source projects.)
Open-source software has been a godsend for companies that have the expertise needed to integrate those projects into their complicated tech stacks, but most companies need some help managing that process and are willing to pay for it. AWS offers an Atlas-like service, but MongoDB Atlas has clear momentum — in part because the licensing restrictions imposed two years ago limit the functionality of AWS' DocumentDB service.
Whether MongoDB would have enjoyed the same success competing head to head without having put the restriction in place, we'll never know. But Ittycheria believes the popularity of Atlas — which is also available on the cloud platforms provided by Google and Microsoft, as well as AWS — is driving revenue for those providers they wouldn't have otherwise seen. In a recent interview with Protocol, Ittycheria talked about that complicated backstory, how MongoDB thinks of the competitive landscape, and the company's plan to return to its New York offices amidst the ongoing pandemic.
This interview has been lightly edited and condensed for clarity.
One thing that stood out to me from your recent earnings was the performance of Atlas over the last year, how that growth seems to have actually accelerated amid the competitive pressure from cloud providers. Given the performance of Atlas, do you still think you needed to make some of those licensing changes that you did to the core open-source product?
The popularity of Atlas really speaks to just the strong product-market fit of MongoDB. MongoDB is by any objective measure the most popular modern database in the world. It took us about 10 years to get to $100 million in revenue, by 2017. Since then we have essentially become a half a billion dollar run rate business in the last four years.
That's kind of my point, though, is that there clearly has been a market response to Atlas, which is the MongoDB flavor of this managed open-source project. Would that have happened had you not changed the terms of the core open-source project?
I mean, you won't really know, but I would say that a big part of what happened is that a lot of people suggested, "your adoption is going to be hurt because people won't consider this a real open-source product." That's the furthest thing from the truth in terms of what's happening. Our popularity has exploded.
But we do think that was important for us. We roughly spend about 50% of our R&D [budget] on the [core MongoDB open-source project], which is free, right? A lot of cloud providers spend a minuscule amount of R&D on open sourcing their products; 99.999% of their work is on work that they're trying to monetize. And we just did think it was patently unfair for someone to take our hard work, our time and our money and then try and take advantage of that. And so we made that change.
The cloud providers are the beneficiary of our growth in two ways. One is we obviously consume more compute and storage as part of provisioning the Atlas service: As our Atlas business grows, they are a net beneficiary of that growth as well. The second way is that Atlas customers end up using 5x to 7x more of whatever [they spend] with us on other services adjacent to Atlas that the cloud provider themselves provides. If [Atlas customers are] spending $100 with us a month, they could be spending anywhere from $500 to $700 on other services from a particular cloud provider. So they really do benefit from the growth of Atlas, and that's why they've been big partners.
What are the types of businesses or the types of applications that you see pop up most often among your customers?
We really are a general-purpose platform. Customers are using it for offloading workloads from the mainframe, for payments, for gaming, for trading apps on Wall Street, for content management systems, for mobile and IoT. We have utility companies in Europe replatforming off Oracle to basically build a billing platform. We have cutting-edge startups like Coinbase, running their cryptocurrency platform on MongoDB.
What we just offer is a much better way for developers to work. They're moving away from a tabular model to a document model, where it's much easier, much more natural, to express relationships between data.
One of the fascinating things about this space over the last 10 years has been the explosion in different types of databases. Do you think that can last?
I would argue that there's only a few platforms or technologies that really have escape velocity. There's obviously legacy platforms, and I would say on the new side, we're actually one of [the platforms with escape velocity] .
What we see customers doing is they'll have some sort of legacy standard, some relational database, either Oracle or something open source, and then they'll have some modern standard.
When I joined the company five years ago, there were still questions about who's really going to win the NoSQL space. And so naturally, customers were a little gun shy by putting all their eggs in one basket.
That issue is no longer a debate. It's clear that we've had escape velocity, it's clear that we're multiples bigger than our nearest NoSQL competitor. We're bigger and growing faster, we are financially solvent, and we have close to a billion dollars of cash in the bank. So that issue about customers being nervous, all that friction has gone away.
I've been approached by some other database companies to sell themselves — we're not that interested. So it's clear that a lot of other companies are hurting.
Why are they hurting?
Developers don't want to take on the cognitive load of learning all these types of databases; [they think] that after some point [it] becomes diminishing returns. So people want to consolidate their platforms, because it becomes easier to manage, there's more developers that know those platforms, and so on, so forth.
It's not like there's not going to be two databases again, though, right? There's still going to be multiple options that people employ especially as application needs change. We can't assume that what applications need in 2020 is going to be what they need in 2025.
I think everyone has underestimated how big this market is. So there's clearly going to be some room for some second-tier players. The only question is, how much of a dent that they're going to make in the industry?
Customers themselves are saying, "I want to be able to consolidate because it's just very inefficient for me to work with all these point tools." And so, yes, because the market is so big people on the margins will be able to close business, but you just have to ask yourself, how quickly can they grow? And how meaningful of a software franchise can they ever become, just by virtue of the fact that there's other platforms that have become far more popular, and the flywheel effect is happening?
Do you see a need for the company to become a little wider in terms of the products and services that you provide?
There's real evidence of how we're evolving from being a database company to being a data platform company. What that really means is that we don't want to just only handle OLTP [online transaction processing] workloads, we want to be able to basically address a wider variety of use cases.
No database allows you to query both your online data and your offline data. But what people have said is with the data volumes exploding, I can't store everything online because it just becomes so expensive to do, so I want to be able to tier storage. But then if I tier storage and then archive, I can't query easily.
With Atlas Data Lake, you now can run one federated query on both your online data and your offline data to get views. A simple use case could be if you're a financial services customer online, maybe you have three or six months worth of history, but maybe you want to go back to five years of history and say, "Hey, how much money have I been spending with this one restaurant over the last five years?" You literally could get that data presented to you very easily and very quickly.
That's obviously a very simple example. But it makes the point that being able to query data that's both online and offline becomes a really attractive way to leverage data to drive business decisions. That's one example of how we're kind of expanding our use case.
What is MongoDB doing at the moment with respect to its work-from-home strategy?
We had basically notified employees about two weeks ago that all employees will be able to work from home for the rest of the year.
And we did that for a couple reasons. One, we wanted to just take the uncertainty off the table, because obviously, different parts of the world are opening up and different parts of the world have different protocols and criteria by which they will open up. There's a lot of confusion, and employees really want some certainty so that they can plan their lives. A lot of employees have decided to move, especially if they lived in a highly congested city or small apartment, they wanted the opportunity to be able to move and get somewhere where working from home would not be so difficult.
That being said, roughly a third of our employees do want to go back to the office; not necessarily every day, but they do want to have a place to go to. We are working on a return-to-the-office kind of program.
I was wondering if your perspective on this was different given that you're a New York company; most of the companies I've talked to over the last few months have been in Silicon Valley.
Clearly, you've seen firsthand what the devastation has done to our community and to the economy. Fortunately, very, very few of our employees got infected, just a handful. A lot more employees have had a friend or family members [get infected], and in fact we've had some employees lose family members to the virus, which is really sad.
Correction: An earlier version misstated the name of Redis Labs. Corrected June 18, 2020.
- State of AWS, Microsoft Azure, Google Cloud: February 2020 ... ›
- Google's Thomas Kurian on COVID-19 and the cloud battle - Protocol ›
- Pushed by coronavirus, Wall Street moves toward cloud - Protocol ›
- Apple quietly hiring some of the world's best cloud talent - Protocol ›
- HashiCorp's big step toward a central role in the cloud - Protocol ›
- Big data has eaten the world. Now what? - Protocol ›
- How Abu Dhabi hopes to build a better Silicon Valley - Protocol ›
- The Linux Foundation became a force in enterprise tech. Is that a problem? - Protocol ›