Skip to main content

📘 MongoDB Search architecture

MongoDB Search provides powerful findability capabilities to your data collections. A flexible index configuration allows mapping and indexing only the fields needed, or dynamically mapping any and all fields supported.

The Big Picture​

Applications communicate to MongoDB Search through the same mechanism as all other requests: an aggregation pipeline through a MongoDB driver.

big picture

System architecture​

The MongoDB Search mongot process, built on Apache Lucene, interfaces with the mongod database process to create and manage full-text (and vector search) indexes and queries.

The mongot process performs the following tasks:

  • Creates MongoDB Search indexes based on the index definition.
  • Monitors change streams for the current state of the documents.
  • Processes MongoDB Search queries and returns the document IDs and other search metadata for the matching documents to mongod, which then does a full document lookup and returns the results to the client.

system diagram

Changes to a collection via updates, deletes, or additions are eventually consistent, meaning the index is updated independently of changes to the collection in a separate process, asynchronously. The lag between a change made to the database and refelected in a subsequent search is dependent on many factors such as deployment tier and architecture, the complexity of the index mapping, the other changes that are also queued, and the laws of physics.

The MongoDB Atlas platform makes MongoDB Search deployable in either a coupled architecture alongside the database nodes, or on separate dedicated Search Nodes. Dedicated Search Nodes provide separation of concerns, alleviating resource contention. Dedicated Search Nodes are recommended for production workloads.

coupled nodes

Dedicated MongoDB Search Nodes​

dedicated search nodes