Beefing Up A Cloudy NoSQL Database To Ride The AI Wave

To Andrew Davidson, senior vice president of products at MongoDB, the database business operates in an entirely different type of market than traditional software, where vendors might sell their products into one organization after another, eventually reaching a saturation point. They can grow fast, but it’s tough to keep that growth going after a while.

With databases – MongoDB has been a staple in the NoSQL market with document-oriented database since 2007 – it’s about capturing workloads, so the vendor’s business grows as the number of enterprise workloads expands. And in an era of exploding amounts of data, cloud computing, and the accelerated rise of generative AI, the number of workloads will only skyrocket.

“Software in many ways is defining the economy today,” Davidson told journalists in the runup to this week’s MongoDB.local developer conference in New York City. “If software is defining the economy today, then in many ways we’re in a developer-defined economy. Developers make all this possible and they’re picking the software technologies that make up the back end of the software. Databases, for example, to power all these applications and all of these new technologies, whether it’s what we’ve seen over the last couple of years with low code [and] no code or what we’re seeing over the last twelve months, that tipping point moment for AI. All of this will accelerate and democratize the trend of more and more software, more and more application workloads that will be built all over the world.”

MongoDB plans to get its share of this business, which will be important as the company looks to reach that nirvana that is profitability. As we noted here earlier this year, MongoDB, like other database and storage companies, is still waiting to reach that plateau, despite being the dominant NoSQL database company and counting more than 40,800 customers and having $1.8 billion on hand.

A key for the company will be its Atlas cloud database, a fully managed service that runs on Amazon Web Services, Microsoft Azure, and Google Cloud Platform. Atlas revenue in FY 2023 grew 65.4 percent year-over-year, hitting $810.6 million. Subscriptions for all of MongoDB’s businesses jumped 46.7 percent while services rose 541 percent.

Given all that, it’s not surprising that MongoDB at its event this week is pushing out new functionality in Atlas, nor is it surprising that a good portion of that touches on generative AI, which Davidson – like most others in IT – said will be the next big paradigm shift in tech, after mainframes, PCs, client-server, cloud, and mobile. It will rapidly automate software development, from code generation to testing to debugging, all of which will require data-intensive applications and more data, all in more real-time operational use cases demanding continuous availability.

“It’s going to be increasingly possible to take advantage of multimedia use cases, images, video, audio and geospatial and more, and to do interesting, meaningful things with them,” he said. “And all of this is in service of incredible applications that are going to be extremely powerful and valuable in every industry. New user experiences, natural language processing, and more. We see AI as a driver for economic value in the form of new applications in every industry, and we’re going to be part of that trend.”

MongoDB in part will do this through expanding a partnership with Google Cloud that started in 2018, integrating Google Cloud’s two-year-old Vertex AI large-language models (LLMs) and managed machine learning platform with Atlas to accelerate developers’ work creating AI-based software. MongoDB also is introducing Atlas Vector Search, a tool for enterprises to more easily use LLMs, which require that the data they use come in vectors. Rather than specific keywords as in traditional search, vector search uses semantic meaning and similarity between vectors for retrieving data, building sentences from prompts, or generating images. It’s a concept that’s been around for a while, Davidson said.

“But something’s really changed over the last two years or so and that is, while in the past we needed to have hard-core machine learning and data science chops to do any of this, now folks have been able to take advantage of off-the-shelf models – many of which are open source – that can summarize rich source data that could be text or images or video or sound, and it could summarize these models as inference models, summarize that data into what are called ‘vector embeddings,’ which is a numerical representation or summary of that information,” he said.

Organizations are using specialized databases for storing vectors that LLMs can use, but with Vector Search in Atlas, they can use MongoDB’s cloud-based database. In addition, Vector Search – which is in public preview – opens up new workloads to run on Atlas, including text-to-image.

AI also will accelerate application modernization, so MongoDB is working to make it easier for enterprises to move off more traditional relational databases – which remains the vendor’s key competitors – to its NoSQL platforms. Such migrations aren’t easy, requiring sorting through workloads, updating schemas, modernizing code, and rewriting applications. For MongoDB, they’re important. Davidson said about a third of the company’s business comes from app modernization efforts.

MongoDB last year previewed Relational Migrator, a tool to make that move easier and now is making it generally available. The tool analyzes other databases – Oracle, Microsoft SQL Server, MySQL, and PostgreSQL, among others – and automatically recommends a new data schema. It then transforms the data, migrates it to MongoDB Atlas, and generates code to work with modern applications.

MongoDB is stretching that thinking into the future and is working on a SQL query conversation capability for converting code to MongoDB’s query language. In addition, its engineers are investigating ways to more easily convert application software to work with MongoDB Atlas, he said.

MongoDB also is adding in public preview Search Nodes to Atlas, creating a dedicated infrastructure with optimized hardware for scaling search workloads independent of the database. “We’ve realized is we’re being pulled in the direction of servicing more and more mission-critical, at-scale use case for search,” Davidson said. “This is going to pave us a road into much larger scale use cases in the future.”

Atlas Stream Processing, also in preview, is aimed at processing streaming data like that coming from Web browsing and Internet of Things devices. In addition, organizations can now use Atlas Online Archive and Atlas Data Federation to query Azure Blob Storage, something that MongoDB previously had enabled in AWS.

MongoDB is hoping such new capabilities will keep customers on the Atlas cloud platform and attract new ones as the software field continues to evolve with generative AI and other emerging technologies. The company holds about 44 percent of the NoSQL database space today, but there are strong competitors, including AWS DynamoDB and Apache Cassandra.

Sign up to our Newsletter

Featuring highlights, analysis, and stories from the week directly from us to your inbox with nothing in between.
Subscribe now

Be the first to comment

Leave a Reply

Your email address will not be published.


This site uses Akismet to reduce spam. Learn how your comment data is processed.