Does Snowflake have a native SQL Server connector?

Yes. Snowflake Openflow has two generally available SQL Server connectors: one based on Change Tracking, which works on all editions and replicates the current state, and one based on CDC, which captures every change but is not available on Express or Web editions. Both replicate only tables with a primary key.

Is the Weld sync from SQL Server to Snowflake incremental?

Yes, once you switch a table from the default full sync. Merge mode loads new and changed rows using the primary key and a cursor such as updated_at; append mode adds new rows only.

Does Weld support CDC from SQL Server?

Not yet. Weld's docs list SQL Server CDC as under development. Today, tables sync on a schedule with incremental merge or append, and full syncs clear deleted rows.

What happens if a full sync into Snowflake fails?

Nothing changes in the destination. Weld stages the data in an internal Snowflake stage and loads it with one COPY INTO statement, which aborts entirely on error, so the table keeps its previous contents.

Why do my queries fail with lower-case table names?

Snowflake reads unquoted identifiers as upper case, and Weld quotes the names it writes. Quote the names in queries, or create the sync with upper-case tables and columns turned on so you can query them without quotes.

SQL Server to Snowflake

Load Microsoft SQL Server tables into Snowflake and keep them current for reporting. Snowflake now has its own Openflow connectors for SQL Server; Weld syncs the tables you choose incrementally and lands them next to your other sources.

Start for free View setup guide

4.8 on G2

Source

Weld

Warehouse

Checked against Weld and vendor docs on 10 October 2026

Weld · New sync

SourceMicrosoft SQL ServerConnected

Tables to sync

Your own tables, picked when you set up the sync.

Extract
Load
Keep in sync

DestinationSnowflakeReady

mssql schema

mssql.Synced
mssql.Syncing
mssql.Synced
mssql.Synced

Trusted by data and AI teams

Options

Three ways to get SQL Server data into Snowflake

Snowflake's Openflow connectors for SQL Server are generally available and replicate changes continuously, which suits teams already running Openflow. Weld suits a warehouse fed from many sources with one tool and schedule.

Openflow Connector for SQL Server

Best for: Current-state replication on any SQL Server edition
Effort: Medium: Change Tracking on source, an Openflow runtime
Freshness: Polls Change Tracking for net changes
Good to know: Generally available. Replicates only tables with a primary key, soft-deletes rows with _SNOWFLAKE_DELETED, and keeps journal tables indefinitely, so storage grows unless you clean them up.

Openflow Connector for SQL Server (CDC)

Best for: Audit trails that need every change
Effort: Medium to high: SQL Server Agent, a capture instance per table
Freshness: Every change, in commit order
Good to know: Not available on Express or Web editions. The capture job uses source CPU and I/O, and the transaction log cannot be truncated past the capture position.

Weld

Best for: SQL Server next to SaaS sources in one warehouse
Effort: Low: connect both, pick tables, choose a sync mode
Freshness: Incremental syncs on your schedule
Good to know: Merge or append incremental syncs, and full syncs at midnight to clear deleted rows. CDC for SQL Server is listed as coming soon in Weld's docs.

How it works

How to sync Microsoft SQL Server to Snowflake with Weld.
In the app, with the API or an agent.

Setup guide API guide MCP docs

Weld · Data warehouse

Data warehouseSnowflakeConnected

Data locationChosen when you connect

CredentialsService account or key, stored encrypted

Data

What lands in Snowflake
From Microsoft SQL Server to Snowflake.

SQL Server has no fixed schema for Weld to publish, so what lands in Snowflake is your own tables, one per table you select, plus a column Weld adds to every row.

Your selected tables
Each lands with its columns, minus any you removed and with any you chose to hash, in the schema you name for the sync.
_weld_synced
The time Weld synced each row. Use it to check freshness and to drive incremental models.
Tables on full sync
Loaded through an internal stage with a single COPY INTO, so a failed full sync leaves the existing table untouched.

Example query

How many rows reached a table each day this week?
In Snowflake SQL.

Runs in Snowflake, assuming you named the destination schema mssql and left upper-case names off, so names are quoted as the source produced them. Replace orders with one of your tables.

Snowflake · query

1select2  date_trunc('day', "_weld_synced") as synced_day,3  count(*) as rows_synced4from "mssql"."orders"5where "_weld_synced" >= dateadd(day, -7, current_timestamp())6group by 17order by 1 desc

Weld MCP server

Ask your agent.
It queries Snowflake for you.

Connect Claude, Cursor, ChatGPT or your own agent to Weld over MCP. It reads the Microsoft SQL Server context, finds the tables it needs and runs the query in Snowflake.

Set up the MCP server

Claude CodeOpus 5

ConnectedWeld MCP

Which Microsoft SQL Server tables do we have in Snowflake, and which is the largest?

Looking up what Weld has synced from Microsoft SQL Server.

●weld·search_available_references

⎿raw.mssql.*

tables, columns and freshness

●weld·get_reference_schema

⎿columns, types and nullability

●weld·run_query

⎿select count(*) from {{raw.mssql.…}}

submitted

●weld·get_query_status

⎿DONE

●weld·get_query_result

⎿rows returned

✓Answered from Microsoft SQL Server data in Snowflake

>Ask about your Microsoft SQL Server dataMCP

Under the hood

Things to know.
Before you sync Microsoft SQL Server.

Quoted names are case-sensitive in Snowflake: Weld quotes every name it writes, so a table that lands as "orders" must be queried quoted. Turn on Create tables and columns in upper case to query without quotes. The setting is fixed when a sync is created.
Move to key-pair authentication: Snowflake is discontinuing password-only single-factor logins after August 2026. Weld supports key-pair authentication; use an encrypted private key where your policy allows.
Incremental syncs need a key and a cursor: The SQL Server connector runs full syncs by default. Merge mode needs a primary key and a cursor such as updated_at; append mode does not capture updates to existing rows.
Deletes need a full sync: Merge and append do not capture rows deleted in SQL Server. Schedule a full sync at midnight on tables where deletes matter. Weld's SQL Server CDC, which would stream them, is listed as coming soon.

Sources: Weld SQL Server connector docs · Weld SQL Server CDC docs · Weld Snowflake destination docs · Weld managed fields · Snowflake: About Openflow Connector for SQL Server · Snowflake: About Openflow Connector for SQL Server (CDC) · Snowflake: Comparison of Openflow connectors for SQL Server