Percona Announces General Availability of Percona Search for MongoDB®

Integrated search lets teams automatically generate embeddings within their own infrastructure, without vendor lock-in, and manage search alongside their database

Percona, a provider of open source database software, support, and services, today announced the general availability of Percona Search for MongoDB^®, enabling organizations to run full-text, vector and hybrid search within their own infrastructure. The offering combines local-first automatic embedding generation with operator-based deployment, helping teams bring search into production environments on their own terms.

Teams building search capabilities for sensitive business information need control over where that information is processed. But generating embeddings can introduce an external service dependency, with per-token charges that grow alongside data and query volume. For organizations with strict data-handling policies, that dependency can stop a project before it reaches production. Teams must also manage search-index backup, recovery, scaling and monitoring.

Marketing Technology News: MarTech Interview with Gil Bouhnick, Director of Product @ AppsFlyer

Built on MongoDB’s vector, full-text, and hybrid search engine, mongot, and integrated with Percona Server for MongoDB (PSMDB), Percona Search for MongoDB works with the Percona Operator today. Support for Percona Backup for MongoDB (PBM) and Percona Monitoring and Management (PMM) is on the roadmap, so teams will be able to manage the search layer with the same tools they already use for their database. The new tool is designed for teams running self-managed MongoDB environments, especially organizations with strict data-residency requirements, as well as existing Percona users adding search to production applications.

“For many organizations, adding vector search has meant sending data somewhere their policies won’t allow,” said Peter Farkas, CEO of Percona. “That’s where a lot of those projects stop. Search should be held to the same standard of control as the database it runs on. Percona Search for MongoDB keeps embeddings inside the customer’s own environment and runs search freely and without lock-in.”

Marketing Technology News: Segments Are Dead. Segmentation Has Never Mattered More.

Local Search Built for Production

Key features of Percona Search for MongoDB include:

• Local embeddings without per-token service charges: An OpenAI-compatible interface supports local embedding providers including Ollama, vLLM and LocalAI, as well as in-cluster Hugging Face TEI models.
• No embedding model vendor lock-in: Teams that prefer hosted embedding generation can use the Hugging Face Inference API, Azure OpenAI or OpenAI to access more than 2,000 models that best suit their needs.
• Support for air-gapped environments: With local embedding providers configured, document text stays within the cluster, and search operates without outbound network access.
• Operator-managed deployment and scaling: The Percona Operator deploys the search tier alongside the database and automatically scales it independently of the database tier.
• Enterprise security included: LDAP, OIDC, Kerberos, audit logging and encryption at rest are standard in Percona Server for MongoDB at no separate license cost.

“Getting search running is the easy part, but keeping it healthy and compliant in production is where teams get stuck,” said Puneet Kala, Head of Engineering and Product, MongoDB at Percona. “Restoring a search index shouldn’t mean rebuilding it from scratch, and teams should know when search results start to degrade before their users notice. That’s why we’re bringing both into PBM and PMM next, so teams can run search with the same tools they already rely on for their database.”