Today at its user conference MongoDB.local NYC, the popular database company announced that the Search and Vector Search capabilities that have been available in the Atlas cloud platform are now available in preview in the Community Edition and Enterprise Server.
Previously, customers using self-managed versions of MongoDB would have needed to use a third-party service for vector databases, leading to a fragmented search stack that adds unnecessary complexity and risk, according to MongoDB.
Ben Flast, director of product management at MongoDB, explained that the team had been working on bringing this to the Community Edition and Enterprise Server for a while, and have finally gotten to a point where it’s ready to be added.
“We brought Search and Vector Search to market in Atlas only six or seven years ago, and the intention there was really like where did we think we could build a new service and evolve it very quickly, and we felt like a managed software would be an easier place to get that product started and get it to a more mature place. And now that we’re there, we’re really excited to bring it to the community because so much of the way MongoDB is used is in the Community Edition,” he said.
According to Flast, one of the biggest considerations was making sure that Search and Vector Search could be as scalable and performant in self-managed versions as it is in Atlas.
“What we released today is the binary that sits underneath the search capability. By having it as a standalone binary, you can put it on separate hardware, you can scale it up independently or run it locally and have a single instance,” he said.
Vector search unlocks capabilities like autocomplete and fuzzy search, search faceting, internal search tools, AI-powered semantic search, RAG, agents, hybrid search, and text analysis.
According to MongoDB, several of its partners helped to test and validate these search capabilities in the Community Edition, including Volcano Engine Cloud, LangChain, and LlamaIndex.
Updates to Queryable Encryption
MongoDB also announced the latest release of its platform, 8.2. Compared to MongoDB 8.0, the latest version provides 49% faster performance for unindexed queries, 10% faster in-memory reads, 20% faster array traversal, and almost triple the throughput for time-series bulk insertions, according to the company.
MongoDB 8.2 also adds partial match support to Queryable Encryption technology. MongoDB explained that this allows text searches to be done on encrypted data without revealing the underlying information.
Queryable Encryption allows data to be protected at rest, in transit, and in use. According to the company, encryption at rest and in transit is commonplace, but encrypting data that is in use has been harder to achieve. This is because encryption makes data unreadable, and queries cannot be performed in this state.
“For instance, a healthcare provider may need to find all patients with diagnoses that include the word ‘diabetes.’ However, without decrypting the medical records, the database cannot search for that term,” the company wrote in a blog post. To work around this, organizations often leave sensitive fields unencrypted or build separate search indexes.
With Queryable Encryption, queries can be done on the encrypted sensitive data without that data ever being exposed to the database server.
MongoDB MCP Server
After a successful public preview, MongoDB announced that its MCP Server is now generally available.
As part of today’s release, enterprise-grade authentication with OIDC, LDAP, and Kerberos has been added, along with proxy connectivity. There is also now self-hosted remote deployment support so that teams can share deployments and have a centralized configuration.
The MongoDB Server can be obtained in a bundle with MongoDB for VS Code extension.
MongoDB AMP
Additionally, yesterday, the company announced MongoDB AMP, a platform that applies AI to the application modernization process. MongoDB AMP consists of an AI-powered software platform, delivery framework, and experienced engineers to guide the technical implementation process.
Shilpa Kolhar, SVP of product and engineering at MongoDB, explained that the AI agents will tackle tasks like adding documentation that was missing or adding functional tests, and then experts can take over when situations arise that the tooling can’t handle on its own.
“When you are converting from your legacy Java stack to Java Spring Boot, it’s a standard framework. The tools handle most of it and the customers see a huge reduction in time for code transformation. But it’s not just about code transformation, right? We want to have the code transformation in place and follow all the best practices that are needed in application development. And companies might have specific needs for their security and compliance, and so on, and that’s where our experts come in,” she said.
She explained that many times, customers will come in and say they have one database, but then the transformation begins and they discover they have many different ones. “That’s where we need to bring multiple tools together, and that’s another area where our experts come in and tie various things together,” she said.
According to Kolhar, this possibility for such varied infrastructure is one of the things that makes legacy systems such a problem. Once an organizations goes through the modernization process, however, their infrastructure will hopefully be standardized in such a way that future changes become much simpler.
She also explained that for a while, there’s been a back and forth of companies putting off modernization because they can’t guarantee the return on investment, but we’ve reached a point in time now where legacy databases and application platforms can’t keep up with the pace AI is changing things.
She also said that because of automation, modernization can happen much faster, and not as many people need to be dedicated to the process.
“We are ready to help you with the tooling we have built over the last few years and the experience of the last 15 years,” she said.
