Database CI/CD overview
Production releases often handle database changes and application deployment separately. A DBA may need to approve the database changes, while existing CI/CD tools deploy the application. CloudDM connects the two so teams can check, approve, and execute database changes before continuing the application release.
- Trigger database changes through a webhook, remote request, or the console.
- After executing a database change, CloudDM can send a callback so the external CI/CD system can continue the release.
- SQL can go through security checks and ticket approval before execution.
Core concepts
- A change flow defines the change source, target database, trigger, execution policy, and notifications for ongoing database changes.
- A change is the smallest unit of database CI/CD execution. An ordinary change goes through submission, SQL review, ticket approval, and execution. CloudDM can close a change when it finds no new SQL. Building a baseline follows the snapshot change path.
How it works
A change flow applies new SQL from a source repository to a target database. Once triggered, CloudDM processes the change in four stages.
- Submit retrieves SQL from the repository and compares it with the existing snapshot to identify new SQL.
- SQL Review checks the new SQL against the security rule assigned to the target database's environment. See Assign a security rule to an environment.
- Work Order Approval starts the change ticket workflow assigned to the target database's environment. See Ticket settings.
- Execute applies the approved SQL to the target database according to the change flow's execution policy.
Limits
- SQL repositories can use GitLab Self-Managed or Gitee.
- A repository can contain multiple SQL files. CloudDM sorts them by path and file name to determine execution order.
- CloudDM processes files with the
.sqlextension. - Append new SQL to the end of an existing or new
.sqlfile. CloudDM detects additions at the end of the file. - Each change flow can have one change in progress at a time.
- Only one enabled change flow can use a given repository, branch, target DataSource, and database or schema combination.