Integrations
Pull data into MDM on a schedule from any Snowflake table, and publish your golden records as governed views that the rest of your Snowflake ecosystem can query directly.
Two directions, one page
The Integrations page covers both halves of the data exchange between Rapida MDM 2.0 and your wider Snowflake environment:
- Inbound syncs — scheduled pulls from any Snowflake table into an MDM table, landing as governed records.
- Outbound Golden Record views — flat Snowflake views exposing your approved masters to downstream systems and analysts.
Active Inbound Syncs
The first tab lists every configured pipeline, one row each:
| Column | What it shows |
|---|---|
| Connection Map | The source location (db.schema.table) and the MDM table it feeds. A WHERE pill appears if the sync has a filter. |
| Schedule | Human-readable schedule — Daily at 14:30 UTC, Weekly at… (Monday), or Manual. |
| Last Sync | When the integration last ran, or Never. |
| Status | Synced or Pending. |
| Actions | Sync Now triggers an immediate run; Delete removes the configuration. |
Setting up an inbound sync
The second tab is a wizard that walks you through the configuration:
-
Choose the target MDM table
Pick the data model and the MDM table the data should land in.
-
Point at the Snowflake source
Provide the source Database, Schema, and Table, then click Verify Connection & Load Columns to test permissions and read the source schema. The application sees only what your Snowflake role can see: no privilege escalation.
-
Optionally add new attributes from the source
If the source has columns that do not yet exist as MDM attributes, select them in the multi-select and they will be created for you, with types inferred from the Snowflake column types (see the mapping table below). Source-level
NOT NULLandUNIQUEconstraints are mirrored onto the new attributes. -
Map the columns
For each MDM attribute on the target table, pick the corresponding source column from a dropdown. Business keys are flagged with a BK badge so you cannot miss them.
-
Add an optional SQL filter
Restrict what the sync pulls with a
WHEREclause, for exampleCOUNTRY = 'Japan' AND IS_ACTIVE = TRUE. -
Set the schedule (UTC)
Choose Manual (runs only via Sync Now), Daily at a specific time, Weekly on a day of the week, or Monthly on a day of the month. Click Save Integration: the configuration is stored and a corresponding Snowflake Task is created or updated behind the scenes.
Type inference from Snowflake
| Snowflake type | MDM attribute type |
|---|---|
INTEGER, BIGINT | Integer |
NUMBER, DECIMAL, FLOAT | Number |
DATE, TIMESTAMP, TIME | Date |
BOOLEAN | Boolean |
| Anything else | Text |
Golden Record Views (outbound)
Every active MDM table is exposed as a flat Snowflake view named
MDM_ONE.CORE.VW_<table>, containing only golden records. This is the
intended way for downstream consumers (dashboards, applications, dbt models, other Snowflake workloads) to read your master data.
The third tab lists every table with:
- The view name and a status pill: Active or View Missing (Needs Refresh).
- The number of records currently exposed.
- Expanders for a Data Preview (top 50 records), the Exposed Schema Columns, and a copy-ready SQL snippet showing how to query the view.
- A per-row Refresh View button.
System columns on outbound views
Alongside your own attributes, each view carries a small set of system columns:
| Column | Meaning |
|---|---|
RECORD_ID | The internal MDM identifier. |
CREATED_AT / CREATED_BY | When the record was first created, and by whom. |
UPDATED_AT | The last update timestamp. |
APPROVED_AT / APPROVED_BY | When the record became Golden, and who approved it. |
MDM_ONE.CORE can query the VW_* views directly: no Rapida MDM 2.0 login required. This makes the views the natural
integration boundary between MDM and your data pipelines.
Common questions
Do synced records skip the approval workflow?
No. Inbound syncs follow the same governance as manual entry and file imports: records land in the governed lifecycle and only become golden through approval. The sync automates the arrival of data, not the trust decision.
What permissions does the sync use to read the source table?
The application reads through Snowflake's own access model: it can only see what the executing role can see. The Verify Connection step confirms this before you save, so permission problems surface at setup time, not at 2 a.m.
How do downstream tools consume MDM data: API, export, or something else?
The recommended path is the outbound VW_* views: plain Snowflake views
queryable by any tool with warehouse access (dbt, BI platforms, applications).
No API integration or scheduled export needed; consumers always see the current golden
records.
A view shows "View Missing (Needs Refresh)". What happened?
The underlying table changed, typically a rename or schema change, and the view needs rebuilding. Click Refresh View on the row (or Refresh All Views after widespread changes) and the view is recreated immediately.
What time zone do schedules use?
All schedules are defined and displayed in UTC. Convert from your local time when configuring: a sync set for 14:30 runs at 14:30 UTC regardless of where you are.