Continuous Integration and Delivery for Databases
summary
Claim: CI/CD for databases is practical and necessary, but requires explicit migrations, tooling, and staged rollouts. Demonstration uses Roundhouse to add a PetName column, runs local dev DB migrations, green CI build, and an Azure release that runs migrations as a deployment task via Azure Pipelines, with scripts hashed to prevent accidental re-runs. Alternatives discussed include Redgate SQL Change Automation and EF migrations. Operational lessons: expect long-running index or batch jobs, use dual writes/triggers and batched backfills, and plan multi-step zero‑downtime migrations; one migration demo shows version 50 deployed successfully.
AI-written summary · about
chapters (19)
- 0:00Start
- 5:40Desired State Configuration
- 11:15RedGate SQL Change Automation
- 12:16Roundhouse Migrations
- 34:49Example: Table
- 35:20Example: Column
- 37:31CREATE OR ALTER
- 38:58Renaming a table
- 39:42Splitting a column
- 40:35Splitting a table
- 45:13Dual Writing
- 46:47What about existing data?
- 48:36Update Read Paths
- 49:28Update Write Paths
- 50:42Large Table Migrations
- 54:23Microservices?
- 55:39NoSQL Migration Options
- 57:04Separate transactional from historical
- 58:19Parting Thoughts
related talks
50mBringing DevOps to the Database
45mEnabling Continuous Delivery with Database Practices
01h 00mLet's Get Scary and Continuously Deploy Database Changes!
46mA Database DevOps Pipeline
46mDatabase DevOps - It is Possible!
01h 10mContinuous Integration for the BI Developer
56mDeploying Docker for Database DevOps Dominance
05mMaking Database Changes a Tier one Artifact
