MongoDB Connector
Sync collections from MongoDB Atlas or self-managed MongoDB into your warehouse with cursor-based incremental sync and optional document packing.
Evaluating against Fivetran? See how Supaflow handles MongoDB pricing, connector quality, and Snowflake deployment side-by-side.
Supaflow vs FivetranWhy Supaflow
All connectors included
Every connector is available on every plan. Pricing does not increase with connector count.
Pay for compute, not rows
Credit-based pricing. Usage scales with your pipelines, not with row counts.
One platform
Ingestion, dbt Core transformation, reverse ETL, and orchestration in a single workspace.
Capabilities
Connection string or split credentials
Paste a MongoDB Atlas connection string with embedded credentials, or keep the URI and the credentials in separate fields with a configurable Authentication Database. Passwords are stored encrypted, and the connector refuses to silently merge two sets of credentials.
Per-collection cursor selection
Schema discovery samples documents per collection and surfaces every datetime root field as a cursor candidate. You pick the cursor in the pipeline wizard, so different collections can use different cursor fields and run independently.
Document Packing Mode
Choose how nested document structures land in your destination. By default the connector keeps embedded documents and arrays as JSON values on the parent row (Packed) so warehouse rows mirror Mongo documents one-to-one. Switch to Unpacked to expand embedded documents into parent__leaf columns and surface arrays of objects as related child outputs.
Configurable sample size for schemaless collections
MongoDB documents do not declare a schema, so the connector samples up to a configurable number of documents per collection during discovery. Increase the sample size to improve type coverage on collections with sparse fields.
Skips system databases by default
Implicit discovery automatically skips MongoDB system databases (admin, local, config) and system collections (e.g., system.profile) so they never need to be filtered out manually. Explicit Database Filters can opt a system database back in; system collections inside any discovered database remain skipped unless you name them explicitly via Collection Filters.
Supported Objects
Discovered Objects
Collection
Every readable non-system collection becomes a root object. In your destination it lands as `database.collection` (e.g., `sales.orders`), so collections sharing a name in different databases stay distinct.
Nested array children (Unpacked mode only)
When Document Packing Mode is set to Unpacked, arrays of objects inside a document are surfaced as related child outputs (e.g., `sales.orders__line_items`). The default Packed mode keeps these arrays as JSON columns on the parent instead.
Skipped by Default
admin, local, config
MongoDB system databases are skipped during implicit discovery. They can still be selected explicitly via Database Filters if needed.
system.* collections
MongoDB system collections (e.g., system.profile) are skipped during implicit discovery; they remain selectable through Collection Filters when required.
How It Works
Create or pick a read-only MongoDB user
Use a dedicated service user with read access on every database you want to sync. For long-term stability, do not use an individual person's credentials.
Copy the connection string
In MongoDB Atlas, click Connect on the cluster and copy the standard or SRV connection string. For self-managed deployments, build the URI with your hosts, replica set name, and TLS options.
Enter the connection string in Supaflow
Paste the URI into MongoDB Connection String. Either embed credentials in the URI, or keep the URI clean and use the Username, Password, and Authentication Database fields. Credentials are stored encrypted.
Optionally narrow the scope
Leave Database Filters and Collection Filters blank to discover everything readable, or list specific databases (e.g., `sales, crm`) and fully qualified collections (e.g., `sales.orders, crm.contacts`).
Test and save
Click Test & Save. Supaflow validates the URI with a lightweight ping, then runs schema discovery for each selected collection.
Use Cases
Operational analytics on multi-tenant MongoDB
When each tenant lives in its own MongoDB database, a single Supaflow source discovers all of them and lands tenants side-by-side in the warehouse with database-qualified table names ready for SQL joins.
Order, customer, and event reporting from product MongoDB
Land core collections like orders, customers, and events into your warehouse, then build dbt models on top. Optionally switch Document Packing Mode to Unpacked so nested line items, addresses, and other arrays of objects surface as related child tables.
Replace ad-hoc Mongo exports
Replace one-off `mongoexport` scripts and Lambda jobs with a managed pipeline that handles incremental cursors, schema drift, and credentials rotation in one place.
Frequently Asked Questions
Does Supaflow support MongoDB Atlas, self-managed MongoDB, or both?
How does Supaflow handle multiple databases on the same cluster?
How does the connector decide what becomes a child object?
How is incremental sync set up?
What permissions does the MongoDB user need?
Does Supaflow support MongoDB change streams or oplog tailing?
Will my schemas remain stable across runs?
Related Resources
Need a connector we don't support yet?
Build one with AI-powered Connector Dev Skills.
Learn More About the Connector SDK
