Skip to main content
This page covers what your Google Cloud administrator sets up. The steps you take inside Inflection are in Import from Data Warehouse and Export to Data Warehouse.

​
Prerequisites

  • Permission to grant IAM roles on your Google Cloud project or on the BigQuery datasets involved.
  • For import, source tables or views that meet the source table requirements in Import from Data Warehouse.
BigQuery authenticates through IAM grants to an Inflection-managed service account. You never enter a key or password in Inflection.

​
Connection Fields

Import connection Export connection

​
Network Access

None. BigQuery does not require an IP allowlist.

​
Prepare BigQuery for Import

Inflection reads your tables directly from your dataset. It keeps nothing in your project and needs no write access. The Connect step of the import wizard shows a Required permissions panel with the Inflection service account to share with. Copy the principal from there.
  1. In the Google Cloud console, select the project that holds your BigQuery datasets.
  2. Open IAM & Admin and click Grant Access.
  3. Add the Inflection service account shown in the wizard as the principal.
  4. Grant it read access to the tables you want to sync. The wizard lists the permissions as bigquery.tables.get and bigquery.tables.getData. The simplest way to grant them is the predefined BigQuery Data Viewer role on each source dataset, which also lets Inflection list the tables in the dataset:
To limit access to specific datasets instead of the whole project, grant the role on each dataset in the BigQuery console under Sharing → Permissions. Back in the wizard, click Test connection, then continue to Step 2. The Warehouse access chip confirms which datasets Inflection can read.
Inflection cannot create objects in your BigQuery project, so Create it for me is not offered when Inflection drafts a view. Click Copy SQL, run the statement in your dataset, and click Verify View.

​
Prepare BigQuery for Export

Inflection creates and maintains the export tables itself, so there is no table-creation script for BigQuery. Grant the write service account the permissions it needs on the destination dataset.
  1. In the Google Cloud console, select the project that holds the destination dataset.
  2. Open IAM & Admin and click Grant Access.
  3. Add the following principal: inflection-datawarehouse-write@silken-fortress-380708.iam.gserviceaccount.com
  4. Grant a role that includes these permissions on the destination dataset:
The predefined BigQuery Data Editor role on the dataset plus BigQuery Job User on the project covers them. Export to BigQuery connection created and pending activation Once the permissions are in place and the connection is saved, the Inflection team validates the setup and activates the sync.

​
Troubleshooting

Check whether the source column is BIGNUMERIC rather than NUMERIC. Large numeric types can fail the type check in a way that is not obvious from the error. Confirm the exact column type with whoever manages the schema, and cast it in a view if needed.
A row ID based only on user_id is not unique when one person belongs to more than one workspace. Include the workspace in the identifier, for example user_id combined with workspace_id, so each user-workspace pair is a distinct row, and sync the Product user organization object.
An attribute reaches Inflection only if it lives on the table you mapped to Organization. If plan type is a property of a workspace but you mapped a parent company table to Organization, map the workspace table instead.
Yes, once none of its columns are mapped to a field in Unified Mapping. Open the connection and delete it. If mapped fields remain, the delete is blocked and the fields are listed; remove the connection as a source from them first.