“A database is the heart of any application, because data is the heart of any application. Think about Amazon: it’s a pretty UI over a database. A database is where you store information, where you process information, and where you analyse information.”
MongoDB CEO Dev Ittycheria is getting back to basics with Computer Business Review, as we wrap up an interview in NYC; behind the boardroom door of the midtown Hilton, a crowd of enthusiastic, polyglottal developers is spending the last day of the company’s conference learning new tricks on its latest core database MongoDB 4.2.
He doesn’t need to finish the sentence: more information/data is being generated than ever. Those that cannot use it – nor build apps powered by it – face getting shut out of their rapidly changing market; amid data-driven disruption that intelligence firm IDC estimates will power the database software market – or the “data management software market” – to an annual value of $98 billion in 2023, up from $64 billion in 2019.
“If we get to five percent share [MongoDB] would be a $5 billion revenue company,” he suggests, cautioning that this is not an explicit company target. “We have north of 50 percent of the Fortune 100 companies as customers. But in terms of wallet share we have maybe two to three percent on average of those wallets. We have a land and expand business; we can grow those accounts for many many years to come…”
Good Relations: New Applications
The opportunity is clearly huge.
Here’s why: the database world has long been dominated by relational databases like those provided by Oracle, in which you pre-define your database schema based on your business’s requirements and set up rules to govern the relationships between fields in your tables.
In such databases, related information is stored in separate tables, then linked through the use of foreign keys and joins. While such databases are fast, powerful and deeply entrenched in many businesses, any changes in schema require a migration procedure that can take the database offline or significantly reduce application performance; both unpalatable, at best, in today’s enterprise or organisational environment.
Many companies running relational databases are replacing, or augmenting, them as a result with more flexible non-relational databases. They have no shortage of choices, but MongoDB is making a strong showing: not least being ranked the “most wanted database” for consecutive years in the Stack Overflow Developer Survey, the world’s largest developer survey, with over 100,000 respondents.
Among those that have turned to MongoDB and its MQL query language to help them do so: Adobe, AstraZeneca, Barclays, Cisco, eBay, Google, SAP, and more; a customer base and success story that has catapulted it, post-2017 Initial Public Offering, to a market capitalisation of over $9 billion, on revenues of circa $350 million.
“Duplicating Data Used to Cost a Fortune”
Ittycheria, an entrepreneur and venture capitalist who sold BladeLogic – the data centre automation specialist he co-founded – to BMC for $800 million in 2008, explains the shift: “When the relational database came out [in the 70s], and even until the late 90s, storage was incredibly expensive; people were very nervous about duplicating data because they knew it was going to cost them a fortune.”
“The way [people avoided this problem] was by using data tables and joins to find information; that way you are very efficient in your usage of a disc. Fast-forward 40 years later, storage and compute are going to zero asymptotically; look at S3 today, it is so cheap. The biggest constraint today [instead] is developer productivity.”
What company today says ‘you know what? I’m innovating too fast’?
“It’s all about innovation speed. You do that by having a lot of development capacity. MongoDB’s become so popular because we really make developers more productive…”
Not Without Hiccups…
The company’s rise in popularity has not come without hiccups.
MongoDB, ostensibly an open source company, has faced issues ubiquitous to those building companies around open source codebases: public cloud providers taking their IP and creating their own managed services; users giving the company a bad reputation by downloading, configuring and running free MongoDB database software without paying the slightest attention to security (an exposed database with default credentials to a Chinese nuclear plant was one recent eye-popping find) and more.
Computer Business Review puts the challenges to the MongoDB CEO, starting with the open source issue. (MongoDB has tackled the “AWS problem” with a new SSPL licence, which requires those using services based on its code to open-source all programs used to make the software available as a service. It’s a serious disincentive to run managed open source variants, and one that saw Red Hat drop MongoDB support from Red Hat Enterprise Linux, saying “the SSPL is intentionally crafted to be aggressively discriminatory towards a specific class of users.”)
MongoDB CEO: Open Source Licence Not a Concern
MongoDB CEO Dev Ittycheria bats the issue aside: “Our business had only grown faster since then. This has made no impact. It only impacts people who may be thinking about taking our free version and offering it as a managed service to third parties.”
He adds: “We’re very committed to the open source communities and building a free product that people use. [But] what we don’t think is reasonable is for a cloud vendor to come and take a free version, monetise and not give anything back.”
“The difference is that the reason we can kind of control on licensing unlike other open source companies is most other open source companies were built on the backs of some prior art. You saw that with Yahoo! spinning out Hadoop; Facebook built Cassandra…”
“The big difference is that those decisions were made to basically get the community to do crowdsourced R&D: say ‘hey I built something interesting; it’s not really core to my business, so we’ve put in the public domain’. By definition the licence has to be very permissive because you want to encourage people to develop it and make it better.”
“But MongoDB was built by MongoDB. There was no prior art.”
“So one: it speaks to the technical acumen of the team here. And two: we didn’t open source it to get help from the community, to make the product better. We open sourced as a freemium strategy; to drive adoption.”
Misgivings over Misconfigurations?
With regard to misconfigured MongoDB databases spewing Chinese nuclear secrets (or other less toxic data onlin;, is that not a reputational risk?
Unlike, say, Datastax, a specialist Apache Cassandra managed services provider, MongoDB’s company name is identical to the name of the product on its bottom rung: the free, open source database.
Ittycheria says customers understand the issue: “The security protocols on our managed services are very, very strong. The configuration issues you hear about are someone downloading free software, building an application, putting some data in and leaving the front door open.”
“We sell to a pretty informed customer. They can quickly tell if something was a technology problem or a configuration problem. We have large banks running mission critical applications on MongoDB; it’s not that security features aren’t available.”
He admits: “Out of the box we’ve made the defaults more strict. But the challenge of being an open source company is that anyone can download your software: a 15-year-old building an app in China, in Russia, in the US. That’s why we encourage people to use Atlas (our managed service). At the lowest level it costs less than a cup of coffee per day.”
How Concerned are Customers About Vendor Lock-In?
“This is a theme that customers bring up a lot with us.”
“One of the real value value proposition of MongoDB to be is that you can run it anywhere. You can can start writing your app on your laptop; you can run on your laptop, in a data center, you can run it in your private cloud and you can also run hybrid environment and you can run it on any cloud provider. You get complete optionality. For a customer it means you can future proof your application.”
“If you don’t like our product, just use the open source version. So you get full and complete optionality and peace of mind.”
How Different from Apache Cassandra?
While Hadoop’s decline has been much-chronicled, another form of database, Apache Cassandra is increasingly popular. Is it a threat? It’s like comparing apples and oranges, Ittycheria explains: “A simple example. If Computer Business Review had an employee database and used something like Cassandra, you have to decide what your primary key is: it could be last name; it could be location. It could be department.”
“Cassandra was optimised for blazingly fast performance. But they force you to make some trade-offs because you can only create the data off your primary key. The challenge with that approach, is that you want to create different ways [to store data]. Today it may be my last name; tomorrow maybe by ethnicity; the day after maybe by location, and so on: because you want to understand your employee population -what’s happening maybe by age and and other diversity statistics.”
“With MongoDB we took all the benefits of SQL and embedded that with our query language. You can query it the same exact way, but you get all the benefits of the modern database and very high performance, scalable distributed architecture; so you can match massive amounts of data. That’s a reason we’re viewed as a general core system for many companies like utility companies who’ve built their the billing systems on us, or among any banks, or retailers building large e-commerce transactional systems. Cassandra is typically more of an adjunct.”
And with that he’s off to talk to customers: “We’re peeling off workloads.”
“People don’t need to throw Oracle out of the door. We’re building confidence and when customers see how much easier it is, how much faster they can deliver value to their users, then they use MongoDB for more and more things.”