change management

Technical Article

Stairway to Reliable Database Deployments Level 6 – Managing Changesets Over Time

  • Stairway Step

As database changesets accumulate over time, they gradually become more than deployment artifacts. This level explores how a well-organized changeset library evolves into a replayable representation of the database’s structural history, supporting not only reliable deployments but also reconstruction, auditing, and long-term maintainability. Through practical examples and a real-world case study, it shows why maintaining coherent Create and Rollback scripts continues to provide value long after a successful production deployment.

You rated this post out of 5. Change rating

2026-08-05

641 reads

Stairway icons Database Deployments

Concurrency and Baseline Control: Level 5 of the Stairway to Reliable Database Deployments

  • Stairway Step

As database development scales across multiple contributors, concurrency becomes an inherent aspect of the process. Changes that are individually correct may still conflict when developed against a shifting baseline. This level introduces a disciplined approach to managing parallel changesets by controlling their scope, identifying overlap, and resolving conflicts through realignment. By ensuring that only one changeset retains its original baseline while others are adapted accordingly, deployments remain predictable and consistent. Once released, changesets are treated as immutable, and further evolution proceeds through new, forward-only changes.

You rated this post out of 5. Change rating

2026-06-24

1,141 reads

Stairway icons Database Deployments

Stairway to Reliable Database Deployments Level 2 - Defining the Deployment Contract

  • Stairway Step

Level 2 formalizes the behavioral guarantees that a changeset must provide in order to be safely deployed and rolled back. It introduces the deployment contract, checkpoint semantics, and the structural scope of Create and Rollback scripts. Data changes are addressed through a dedicated Update mechanism, with clear boundaries and limitations. By the end of this level, a changeset becomes a predictable and well-defined unit that can be reasoned about independently of execution context.

You rated this post out of 5. Change rating

2026-06-11 (first published: )

1,220 reads

Blogs

Exploring DiskANN: Part 2: PQ, SSDs, caching and beam search

By

How I used AI for this postChatGPT to generate images based on info specifically...

T-SQL Tuesday #202: The Two Outages I Won't Ever Forget

By

My T-SQL Tuesday #202 entry for Marlon Ribunal's invitation on memorable SQL Server outages....

Measuring RAG Solutions: Are We Retrieving the Right Information?

By

A RAG pipeline that answers questions in the demo is not the same thing...

Read the latest Blogs

Forums

Performance Regression After Upgrading from SQL Server 2016 to 2022

By abdalah.mehdoini

Hello everyone, Last week we performed an upgrade of our SQL Server instance, moving...

Adding new column with DEFAULT

By Thomas Franz

Comments posted to this topic are about the item Adding new column with DEFAULT

How do I connect to a SQL Server database on my ISP?

By Doctor Who 2

I've got a web application for my side business. I've added a SQL Server...

Visit the forum

Question of the Day

Adding new column with DEFAULT

Which number will the COUNT() return after executing the following statements:

DROP TABLE IF EXISTS #test;
CREATE TABLE #test (id INT)
INSERT INTO #test (id)
SELECT *
  FROM GENERATE_SERIES(1, 3) AS gs
;

ALTER TABLE #test ADD flag BIT CONSTRAINT DF_#test_flag DEFAULT 0;
go
UPDATE #test SET flag = 0 WHERE id = 1
UPDATE #test SET flag = 1 WHERE id = 2

INSERT INTO #test (id) VALUES (4)

SELECT COUNT(*)
  FROM #test AS t
 WHERE flag = 0
 

See possible answers