Question 1

How does writing into MongoDB work?

Accepted Answer

As a model sync with mapped document fields and paths rather than bulk table replication, since MongoDB is not available as a bulk sync destination. You decide the document shape and map columns onto it, including nested paths. The collection can be created for you and arbitrary fields added, so the destination structure does not need to exist in full before the first run.

Question 2

Do nested Databricks values survive?

Accepted Answer

They map across more directly than in most destinations, since both sides model nesting properly. STRUCT and ARRAY retain their nested schema where it can be parsed, and MAP arrives as a generic JSON shape, all of which can be mapped onto document paths rather than being resolved into columns first. That suits cases where the lakehouse model already has the shape the application wants. Where it does not, flattening in the model SQL remains an option, so the choice sits with you.

Question 3

How are existing documents matched?

Accepted Answer

By the document identifier where you map one, which is the most reliable route, and serialized ObjectID values are converted rather than treated as plain strings. Without one, matching falls back to whichever destination field you nominated as the identity. Inserts and updates are grouped into bulk write calls, and classification into insert or update work happens before the write rather than relying on an upsert flag.

Question 4

Is there anything to watch on types?

Accepted Answer

Two things. Datetimes are strictly coerced, so a malformed value becomes a record level error rather than being written as a string, which surfaces problem rows rather than hiding them. Worth knowing too that TIMESTAMP is timezone aware and normalised as an instant while TIMESTAMP_NTZ is wall clock time carrying no zone, so the two behave differently once stored. DECIMAL keeps precision and scale, with scale zero values sometimes mapping to an integer type where that is safe.

Question 5

How do I connect Databricks to MongoDB?

Accepted Answer

Connect Databricks with a SQL warehouse hostname and a token, or an OAuth service principal, and connect MongoDB with an Atlas connection string or explicit host details. Because this runs as a model sync, the next step is deciding the document shape rather than picking a table to copy. Map columns onto document fields and paths, map an identifier so existing documents are matched, then set a schedule.

Connect Databricks with MongoDB

Connect Databricks and MongoDB with Polytomic

Nested values fit naturally

You define the document

Matches on document ID

Trusted by businesses worldwide

Step-by-step integration guide

Connect Databricks in Polytomic

Connect MongoDB in Polytomic

Create a sync between Databricks and MongoDB

Map fields and enable the sync

Frequently asked questions

The sync and ETL platform that does it all

Powerful data sync

Advanced analytics

Enterprise security

Easy setup

Ready to connect Databricks with MongoDB?