Deep Dive into Declarative Migrations
Prepared for an Atlas Community Webinar, October 2024
Introduction
In recent years, the shift to declarative resource management has transformed modern infrastructure practices. Groundbreaking projects like Terraform, for infrastructure as code, and Kubernetes, for container orchestration, have exemplified the power of this approach. By focusing on what the end state should be rather than how to achieve it, declarative methods make systems more scalable, predictable, and easier to maintain—essential qualities for handling today's complex environments.
However, when it comes to managing database schemas, the industry has been slow to adopt declarative workflows. Atlas was created almost four years ago to address this gap.
Atlas supports two kinds of database schema migration workflows:
-
Versioned Migrations - each change to the database is described as a migration script, essentially a SQL file containing the SQL commands to apply the change. Migrations are versioned and applied in order.
Contrary to most existing migration tools, Atlas relies on users defining the desired state of the database schema in code Atlas generates the necessary migration scripts to bring the database to the desired state.
-
Declarative Migrations - the database schema is described in a declarative way, and changes are applied by comparing the desired schema with the current schema and generating the necessary SQL commands to bring the database to the desired state.
To date, most teams that used Atlas in production have used it's versioned migration workflow which synthesizes the simplicity and explicitness of classic migration tools with the benefit of automated migration generation.
Recent improvements to Atlas have addressed many of the challenges and concerns teams have expressed around using declarative migrations in production in the past. In this post, we'll take a deep dive into the declarative migration workflow
