Question 1

What permissions does the sync account need?

Accepted Answer

For reads, select on the tables you want plus the ability to inspect schema metadata. For destination writes, rights to create and alter tables in the target schema. If you enable the Azure Blob fast path there is more: the database user needs permission to create and drop database scoped credentials and external data sources, and to run BULK INSERT. That last set is not currently listed in the permissions documentation, so it is worth granting deliberately rather than discovering it on the first failed load.

Question 2

How do incremental reads work for Azure SQL?

Accepted Answer

Through a tracking field. You nominate a monotonically increasing column such as an updated timestamp, and each ETL run reads rows beyond the last saved position, with the first run performing a full read. There is no change data capture on this connector, which means deletes are not detected and a row edited without its tracking value advancing will be missed. If deletes matter for your reporting, schedule a periodic full resync to reconcile.

Question 3

What does the Azure Blob storage fast path do?

Accepted Answer

It changes how destination writes are loaded. Instead of batched SQL inserts, Polytomic uploads the data as CSV to Azure Blob using a storage account key, creates a short lived shared access signature, sets up temporary database scoped credentials and an external data source, runs BULK INSERT, and then drops the objects it created. For large loads this is considerably faster. Two things to know: it authenticates with a storage account key rather than a managed identity, and the database user needs rights to create those temporary objects and run BULK INSERT.

Question 4

Can I use Microsoft Entra ID or a managed identity to connect?

Accepted Answer

Not currently. The connector supports SQL authentication only, with no Entra ID, service principal, managed identity, or access token option. For a private database an SSH tunnel is available, and network access through firewall rules or private endpoints remains on your side. On encryption: with SSL enabled the connection is encrypted, but the server certificate is trusted without hostname or certificate authority verification, so traffic is protected in transit without full certificate validation.

Question 5

Can I run reverse ETL into Azure SQL?

Accepted Answer

Yes, and reverse ETL works alongside ETL in the other direction on one platform. Model sync writes support create only, update only, update or create, and replace, all requiring an identity mapping. Bulk replication into Azure SQL merges through staging tables where keys and incremental input allow, propagates soft deletes by default with optional hard deletes, and creates schemas and tables as needed. Schema evolution adds new columns and archives a column under a type suffixed name when its type changes incompatibly, so a sync keeps running rather than failing.

Azure SQL

Perfect pairings

Bi-directional syncing

Blob storage fast path

Automatic schema handling

Example workflows

What you can sync

From Azure SQL

To Azure SQL

How it works

How to set up

Frequently asked questions

Why teams choose Polytomic

No engineering required

Flexible data modeling

Handles scale automatically

Start syncing your data with Azure SQL in minutes

More integrations beyond Azure SQL