Introduction to the MongoDB Connector¶

MongoDB is a document database. Instead of tables with fixed columns, it stores JSON-like documents in collections, and two documents in the same collection can hold different fields, nested sub-documents, and arrays. A MongoDB deployment can be a single server, a replica set, or a cluster hosted on MongoDB Atlas.

The Qualytics MongoDB connector reads the collections of a MongoDB database directly from the server. A Sync turns each collection into a container and works out its fields from a sample of its documents, so Qualytics can profile collections, run scheduled scans, and surface record- and schema-level anomalies on document data the same way it does on tables.

Notes¶

What becomes a container. Each collection in the selected database becomes a container, including time-series collections. Views and the system. collections are not listed. Each database you select in the form becomes its own datastore. The server's own admin, config, and local databases are never offered.

How fields are worked out. MongoDB has no fixed schema to read, so each Sync samples up to 10,000 documents per collection and builds the container's fields from what it finds. A collection with 10,000 documents or fewer is read in full. The fields are then kept as they are until the next Sync, so a profile or a scan always reads the same fields, even when new documents add a field the sample did not see. Run a Sync again to pick up fields that appeared since.

Repeatable samples. When a collection's _id values are ObjectIds, numbers, or dates, the sample is drawn through the _id index from the same points every time, so the same documents give the same fields on every Sync. When _id values are text, mixed, or compound, or the collection is a time-series collection, the sample is random and can differ between Syncs.

Nested documents and arrays. Sub-documents become nested fields and arrays become array fields, profiled and checked as described in Complex Data Types. A sub-document with more than 20 different keys across the sample is read as a map, whose frequent keys are promoted to fields of their own. Field names that contain a dot at the top level, such as a key named a.b, are kept apart from a nested a.b.

Values of an unexpected type. A document can hold a value whose type differs from the one the sample settled on for that field. Where the value converts, it is read as the field's type: the text "12" in a number field reads as 12. Where it does not, only that field reads as empty for that document, and the rest of the document is read normally. The read does not stop.

How MongoDB types are read.

What the form does not ask for. There is no connection string field. You give the host, the port, and the options as separate fields, and Qualytics builds the connection from them, so a password never has to be written into an address. There is no Catalog field either, because MongoDB has no level above the database; Qualytics works it out from the connection settings. The connector signs in with a user and password. Other MongoDB sign-in methods, such as X.509 certificates, LDAP, or Kerberos, are not offered.

Deep Dive¶

How-tos¶

Reference¶