Is converting DuckDB to SQL Server free?

Yes. Files up to 10MB convert free with no account. There is a limit of five conversions per day per IP address.

Will my DuckDB data keep its types in SQL Server?

Yes — that is the point. Whole numbers, decimals, dates and booleans are mapped to real SQL Server types rather than everything becoming text. The mapping table above shows exactly what you get.

Do you store my file?

No. The file is written to a temporary directory, converted, and the directory is deleted as soon as your download has been sent.

Convert DuckDB to SQL Server

Upload a .duckdb or .ddb file and get back a SQL Server SQL script you can load straight into your database. Types are preserved rather than flattened to text. Free, no registration, up to 10MB.

1Upload your file

Click to upload or drag and drop

CSV, Excel, SQLite, SQL, DBF (max 10 MB)

2Choose target format

How your data is typed in SQL Server

Generated from the converter's own SQL Server type mapping — this is what it actually emits, not an example.

Your data	SQL Server type
Whole number	BIGINT
Decimal (money)	DECIMAL(10,2)
Floating point	FLOAT
True / false	BIT
Date	DATE
Date and time	DATETIME2
Short text	NVARCHAR(255)
Long text	NVARCHAR(MAX)
Binary data	VARBINARY(MAX)

Working from a .duckdb file

DuckDB keeps an entire analytical database in one file, with real types behind it - decimals that are decimals, and integers wide enough to be worth checking before you move them somewhere narrower.

→Only the main schema is read. Tables created in another schema, or in a second database attached at query time, are not part of the conversion.
→Views in main are read as though they were tables, so a view's rows are materialised into the target rather than skipped.
→HUGEINT is the one type with nowhere to go: no target here has a 128-bit integer, so those columns are carried across as text rather than silently truncated.
→A DuckDB process that was killed leaves a .duckdb.wal sidecar holding everything it had not checkpointed yet. Upload the database on its own and it opens with no error and no tables at all - send the .wal file alongside it, or checkpoint before you copy the database out.

How names and bytes are written

mssql is the short name the drivers use - the ODBC DSN, SQLAlchemy's mssql+pyodbc, the container image. What it produces is ordinary T-SQL text, and the details worth checking before you run it are how it writes names and how it writes raw bytes.

→Identifiers are wrapped in square brackets rather than double quotes, and a bracket inside a source name is turned into an underscore before the wrapping goes on: a column named Order]Date is created as [Order_Date], so a bracket in your data can never close a name early.
→The identifier cap is 128 characters, high enough that almost every source name survives untouched - one that Oracle would have to cut at 30 arrives here whole.
→Binary columns are VARBINARY(MAX) and their values are written as bare 0x hex literals, with no quotes around them and no conversion function wrapped over them.

Reading DuckDB

→Types, primary keys and NOT NULL constraints are read from the database.
→The file is opened read-only, so your upload is never modified or upgraded in place.

Writing SQL Server

→String literals carry the N prefix, so non-ASCII text survives whatever collation the database uses.
→Timestamps become DATETIME2. T-SQL’s TIMESTAMP is a row-version counter, not a point in time.
→Identity columns are wrapped in SET IDENTITY_INSERT so explicit key values load.
→Statements are separated with GO in batches.

What you get

A ZIP containing dump.sql, plus a README with the exact command to load it. If anything about your data needed a judgement call, a _warnings.txt explains it.

Load this dump into SQL Server:

  sqlcmd -S HOST -U USER -P PASSWORD -d DATABASE -i dump.sql

Notes:

  - Text columns are NVARCHAR and literals carry the N prefix, so
    non-ASCII data survives regardless of database collation.
  - Timestamps are DATETIME2. T-SQL's TIMESTAMP type is a row
    version, not a point in time.

Questions

Is converting DuckDB to SQL Server free?: Yes. Files up to 10MB convert free with no account. There is a limit of five conversions per day per IP address.
Will my DuckDB data keep its types in SQL Server?: Yes — that is the point. Whole numbers, decimals, dates and booleans are mapped to real SQL Server types rather than everything becoming text. The mapping table above shows exactly what you get.
Do you store my file?: No. The file is written to a temporary directory, converted, and the directory is deleted as soon as your download has been sent.