Skip to content
TwinCommitDocs

SQL Server target

TwinCommit applies changes to SQL Server with set-based MERGE statements from a staged batch and writes the checkpoint in the same transaction. Character columns are created with a binary collation by default, so keys compare by code point exactly as they do in the source.

Supported: SQL Server 2016 and later (compatibility level 130+), Azure SQL Database and Managed Instance.

Configuration

Section titled “Configuration”
target:
  type: sqlserver
  dsn: sqlserver://twincommit:${MSSQL_PASSWORD}@mssqlhost:1433?database=Replica
  collation: Latin1_General_100_BIN2   # default; "default" = database collation

Grants

Section titled “Grants”
CREATE USER twincommit FOR LOGIN twincommit;
ALTER ROLE db_ddladmin ADD MEMBER twincommit;
ALTER ROLE db_datareader ADD MEMBER twincommit;
ALTER ROLE db_datawriter ADD MEMBER twincommit;

Collation

Section titled “Collation”

Under SQL Server’s usual case-insensitive collations, two source keys that differ only in case would collide on one row. Keep the BIN2 default unless you know your keys do not depend on case. Set collation: default to use the database collation.