MongoDB readies its Atlas database service for new workloads

At its MongoDB.local NYC event, MongoDB today announced a slew of product releases and updates. Given the company’s focus on its fully-managed Atlas service, it’s no surprise that the majority of news focuses on that platform, with improved support for AI and semantic search workloads, dedicated search nodes to better enabled search use cases and new capabilities to process streaming data, among others.

Andrew Davidson, MongoDB’s SV of product, told me that this is a continuation of the work the company has been doing on Atlas in recent years. “With Atlas, we can deliver capabilities much more quickly,” he said. “We’re able to add the power of search and time series and drive a wider variety of workload shapes.” He argues that as businesses are forced to do more with fewer resources — all while developers are expected to build more applications and do so faster — expanding Atlas’ capabilities is a natural evolution for MongoDB. “We think that this is totally our moment, because we come in with our developer data platform vision, saying: we want to enable a builder to express the vast majority of the features in the vast majority of their applications with respect to their operational data needs. That’s why we keep investing in all of these key primitives and capabilities,” he explained.

Vector search is maybe the most obvious example here. For companies that want to use large language models (LLMs), translating their data into vectors and storing them is key to customizing foundation models for their needs. In addition, vector search also enables new workloads on Atlas like text-to-image search, for example. “We think that, of course, a developer data platform that specializes in operational data should also be able to then express indexes that let you efficiently query the vector summaries of that data,” said Davidson.

Likewise, stream processing is a capability that hasn’t traditionally been the focus of MongoDB’s document model. For a while now, MongoDB has been offering its Aggregation Framework, which allows developers to perform transformations on a stream of documents that comes out of a database. “We realized, ‘holy moly, that’s a perfect metaphor for being able to conceptualize transformations on a stream coming off Kafka,’” Davidson explained.

Another new feature here is support for querying data in Microsoft Azure Blob Storage with MongoDB Atlas Online Archive and Atlas Data Federation. MongDB previously launched support for AWS. While MongoDB would obviously prefer it if everybody hosted their data in MongoDB, the reality is that most enterprises will continue to use m multiple systems. Atlas Data Federation makes it easy for developers to read and write data from and to Atlas databases and third-party cloud object stores, which then makes it easier for them to generate and combine data streams from multiple data sources to power their applications.

Some of the other new features MongoDB is launching this week include Atlas Search Nodes, which are dedicated nodes for scaling search workloads independent of the database, as well as improvements to how the database handles enterprise-scale time series workloads.

“The new MongoDB Atlas capabilities announced today are in response to the feedback we get from customers all around the world—they love that their teams are able to quickly build and innovate with MongoDB Atlas and want to be able to do even more with it across the enterprise,” said Dev Ittycheria, MongoDB’s president and CEO. “With the new features we’re launching today, we’re further supporting not only customers who are just getting started, but also customers who have the most demanding requirements for functionality, performance, scale, and flexibility so they can unleash the power of software and data to build advanced applications to transform their businesses.”