How it compares
Flyway, DbUp, grate, EF Core migrations, SSDT, Sqitch and Liquibase.
Every tool here is good at something. This is how Badger SQL Migrate's design relates to them, for SQL Server, Azure SQL and PostgreSQL. Feature sets change; check each project for its current state.
| Badger SQL Migrate | Flyway | DbUp | grate | EF Core | SSDT / dacpac | Sqitch | Liquibase | |
|---|---|---|---|---|---|---|---|---|
| Free down / undo | Yes | Paid | No | No | Yes (generated) | Partly (older dacpac) | Yes | Yes (targeted: Pro) |
| Two developers adding changes at once | Yes (named + union merge) | No (same V number) | Partly | Partly | Partly (snapshot conflicts) | Yes | Partly (plan conflicts) | Partly (changelog conflicts) |
| Cryptographic hash of every change | Yes SHA-256 | Partly CRC32 | No | Yes | No | No | Yes SHA-1 | Partly MD5 |
| Certifies the order across machines | Yes hash chain | No | No | No | No | No | Yes | No |
| Same SQL on every machine | Yes | Yes | Yes | Yes | Yes | No (diff at deploy time) | Yes | Yes |
| Rollback without the original checkout | Yes (stored in DB) | No | No | No | No | No | No | No |
| Explicit dependencies between changes | Yes | No | No | No | No | n/a | Yes | No |
| Schema drift detection | Yes | Partly (paid) | No | No | No | Yes | No | Partly (Pro) |
| Deploy lock for concurrent pipelines | Yes | Yes | No | Partly | Yes (EF 9+) | No | Yes | Yes |
| Per-environment config and protection | Yes | Yes | No | Partly | No | Partly | Yes | Yes |
| Every Entra ID flow incl. MFA | Yes | Partly | Partly | Partly | Partly | Partly | No | Partly |
| Always Encrypted key provisioning | Yes | No | No | No | No | Partly | No | No |
| Repeatable scripts (views, procs) | Yes | Yes | No | Yes | No | Yes | No | Yes |
| No runtime to install | Yes single binary | No JVM | library | .NET tool | .NET | SqlPackage | No Perl | No JVM |
What Badger SQL Migrate keeps
- From Flyway: validate, repair, baseline and repeatable scripts; a simple mental model.
- From Sqitch: named changes, a plan file, explicit dependencies, deploy/revert/verify triplets and a hash chain.
- From grate: folders for programmable objects that re-run when they change.
- From DbUp: an embeddable library, so an app can migrate itself at startup.
- From SSDT: knowing when the live schema drifted from what was deployed.
What it fixes
- Version collisions. Changes are named; order lives in a line-based plan merged by git's union driver.
bsm plan checkcatches the rare real conflict (two changes with the same name). - Paid or lossy down. Revert scripts are written by you, required by default, free, and stored with the deploy.
- Weak checksums. SHA-256 over normalized content (CRLF and BOM differences do not matter, everything else does) plus a chain over the order.
- Rollback depending on your checkout. The database holds the revert script it will need.
- Non-deterministic deployments. The SQL that runs is the SQL you reviewed, and a rendered-SQL hash records exactly what ran in each environment.
- SQL Server as a second-class citizen. Native T-SQL batch splitting, a T-SQL linter, Entra ID, Always Encrypted, Azure SQL connection settings, TDS 8.0.
Coming from another tool? See Existing databases for adopting a database that is already in production.