Free hands-on database schema training

Shape. Strengthen. Succeed.

Start your path

From blacksmith vocabulary to running real schema across SQL Server, PostgreSQL, MySQL, and MariaDB — and then operating it like a team. Read the primer, then work the courses in order. Every module pairs a quick concept with a hands-on exercise against a local sandbox: declare your state, quench it to a live database, watch it converge.

Speaking forge A two-minute decoder for SchemaSmith's blacksmith vocabulary. Read this first.

Course 1 · First deployment Six hands-on modules — from why declarative beats migration scripts to bringing an existing database under management. Start here.

Course 2 · Going Deeper Products and deployment guardrails, templates that fan out across schemas and databases, conditional deployment, script tokens, data delivery, and custom properties. Start with product boundaries.

Course 3 · Ship it / operate it The operational arc — team workflow and code review, CI gating, rollback as a first-class move, promoting one artifact across every environment, a capstone that survives a careless rename, and what Monday's deploy does with Friday's hand-made hotfix. Start with what changes when schema-as-code joins a team and a pipeline.

Course 4 · Driving the tools A power-features cookbook — eleven independent recipes for making custom properties and script tokens drive what SchemaSmith does: environment-aware schema, self-enforcing policy, deploy-time queries, traveling assets, generated DDL, surviving a destructive rebuild, conditional data delivery, recyclebin hooks, a data dictionary built from your own metadata, a replica provisioned from the model, and an expensive change rolled out one maintenance window at a time.

Course 5 · Switching to SchemaSmith Coming from Flyway, Liquibase, EF Core, SSDT/DACPAC, or hand-rolled scripts? A tool-by-tool migration cookbook — map your old tool's concepts onto SchemaSmith, then extract your live database into declarative files and walk away from your migration history. Flip to your tool, then adopt a drifted production database one guardrail at a time.

Course 6 · Operations & CI Deploy like an operator — a least-privilege account and the datafix profile, pre-flight diagnostics, DB-free CI schema validation and Extensions governance, runtime validation gates, object-level emergency patching with SchemaShears, the --Validate semantic linter, and changing a busy table without blocking it. Start by deploying like an operator.

Course 7 · SchemaSmith at Scale One schema across a fleet — database-per-tenant fan-out, steering the roster from your config, provisioning at onboarding, previewing and throttling and resuming, diagnosing which tenant broke, parsing the deployment summary report, reading the roster from a registry table, handling the read-only copies and secondaries you don't write to, and shipping the registry in the same product as the fleet it lists. Start with what changes when one schema serves a fleet.

Course 8 · Troubleshooting & Diagnostics Read the tool when a deploy fails — the failure roll-up and phase map, structure-change and index/constraint/FK failures, script-slot and data-delivery errors, the recovery toolkit, the per-engine error channels, and what a log gives away before you share it. Start with what to do when the deploy fails.

Course 9 · SchemaSmith for the Polyglot Shop One team, more than one engine — the same change as native packages across SQL Server, PostgreSQL, and MySQL, where native fidelity diverges, organizing polyglot services, per-service pipelines, PostgreSQL replication readiness, and an independent-release capstone. Start with the three-headed shop.

Course 10 · The Rolling Upgrade One package across a mixed-version fleet — let the engine adapt its own DDL, gate your scripts by version, write two shapes of one query, deploy to the oldest tier, retire the gates, and read the first deploy after a server upgrade — plus a bonus recipe on when to ship seed data as XML. One package from SQL Server 2008 through current, PostgreSQL 12 up, MySQL 5.7.22 / 8.0.23 up, MariaDB 10.2.13 up. Start with the mixed-version farm.

Course 11 · What SchemaSmith Refuses To Do The boundary, and why it is the argument — what a declarative tool converges, what it refuses by name, and what it degrades and reports. Placement and partitioning applied at CREATE, the state your package never described, the drop it declines to perform, the database switch a table can't own, and why unset means unmanaged. Start with the three postures.

Course 12 · Declare it, don't script it Stop scripting the objects a guarded CREATE quietly stops updating. Declare PostgreSQL types and sequences, MySQL and MariaDB events, indexes on tables you don't own, full-text search, temporal history and ledger tables, spatial, XML, columnstore, GiST and GIN indexes, the security layer, then prove an extraction gives back exactly what you had and a re-deploy changes nothing.

What you’ll build

Not toy examples — real work, on SQL Server, PostgreSQL, MySQL, and MariaDB, against a live sandbox.

• Your first schema package: one table, declared once, quenched to SQL Server, PostgreSQL, MySQL, and MariaDB.
• An existing database cast into version-controlled files with SchemaTongs — no hand-written DDL.
• Multi-tenant fan-out: one quench that finds every database and schema and converges them all.
• One package, many environments — conditional deployment, script tokens, and data delivery.
• A real operational loop: code-reviewed schema changes, a CI pipeline, rollback, and zero-downtime promotion across dev, staging, and prod.

What you’ll need

• Docker. The labs spin up a throwaway SQL Server + PostgreSQL + MySQL + MariaDB sandbox — start it, work the lab, tear it down.
• The SchemaSmith CLI on your PATH — schemaquench, schematongs, datatongs, and schemashears. No runtime, no service to configure. Course 1, Module 1 walks the install and confirms the connection.

That’s the whole kit. Read the primer and let’s kindle the forge.