Database Build and Release - SQL In The City - Ernest Hwang
2 likes1,129 views
Ernest Hwang discusses automating database maintenance tasks using Red Gate SQL Source Control, continuous integration with Jenkins, and unit testing with tSQLt. He demonstrates committing database changes to source control, how continuous integration runs builds and tests on each commit, and deploying changes between environments with SQL Compare. Automating these processes saves developers time by eliminating manual scripting and deployments while improving quality by catching errors early in continuous integration builds and unit tests.
Database Build and Release - SQL In The City - Ernest Hwang
- 1. Ernest Hwang Principal Software Engineer, Practice Fusion http://practicefusion.com/careers/
- 2. Who am I? • Ernest Hwang, Principal Software Engineer at Practice Fusion in San Francisco (We’re Hiring!) • C#, .NET Developer since 2001 • SQL Server Developer since 1999 (SQL Server 6.x) • Using Red Gate for Continuous Integration since June 2011
- 4. Why am I here?
- 5. What is this? A “How To” guide for automating your day-to-day maintenance tasks by… • *Easily* Versioning your database using Red Gate SQL Source Control • Using Continuous Integration (via Jenkins) to validate Database builds • Using Jenkins to automate database deployments • Using CI / tSQLt / SQL Test to run unit tests
- 6. Prerequisites • Experience with source control management systems (svn, TFS, git, Hg) • Familiarity with Continuous Integration Products (Jenkins, Team City, Cruise Control) • Awareness of build scripting languages (ant, MSBuild)
- 7. What software is used? • • • • • – – – – • SQL Server 2008 / 2012 Red Gate SQL Source Control Red Gate SQL Compare Red Gate SQL Data Compare Jenkins Continuous Integration Server http://jenkins-ci.org/ Promoted Build Plugin Copy Artifacts Plugin Version Plugin Git Plugin MSBuild
- 8. What is SQL Source Control • Source Control plug in for SQL Server Management Studio • Creates a “snapshot” of the database schema defined by CREATE scripts • Snapshot is consumable by SQL Compare and may be used to compare against the schema of an actual database
- 9. What is SQL Source Control? SQL Source Control is to SSMS as TortoiseSVN is to Windows Explorer
- 10. Demo Environment GitHub Developer Workstation Local Copy of RGDemo database Build and Database Server (Windows Azure VM) Dev, QA, & Prod Databases
- 11. Commiting Changes to Source Control Demo Local Database Changes Commit Changes via SQL Source Control Changes committed to SCM repository
- 12. How does Continuous Integration fit in? • The CI server polls the repository for changes • When changes are checked in, the CI job kicks off – Verifies that the database can be built • Builds a brand new database from scratch using SQL Compare and SQL Data Compare – Runs unit tests • Build should fail if unit tests do not succeed – Archives the artifacts (for deployments) • Artifacts can include build/test reports – Emails engineers if there are problems
- 13. I’M THE ONE WHO DROPS TABLES! (demo)
- 14. Deploying Changes to Different Environments • The Promoted Builds plug in can be used to deploy changes to Integration/QA/Staging/Production environments • SQL Compare and SQL Data Compare are used to deploy changes between the sourced controlled database and your development environments • Deploying to Production and/or Staging can be configured to just create the scripts as opposed to forcing the synchronization
- 15. Deploying to Dev / QA Demo SQL Source Control Repository SQL Compare Development Environment Database
- 16. Creating a Database “Version” • Make sure the Jenkins Versioning plugin is installed • Create a User Defined Function called dbo.DATABASE_VERSION() • Create a build step that updates the UDF with the version number • The updated UDF will be archived and used with deployments
- 17. Versioning your Database Demo Commit Changes via SQL Source Control CI Server Detects Changes Kicks off Build Process Build Task updates dbo.DATABASE_VERSION() Function Updated UDF Archived
- 18. Unit Testing with tSQLt and SQL Test • tSQLt is an open source set of stored procedures and functions to facilitate Unit Testing in SQL databases • SQL Test is a wrapper around this framework that integrates with SSMS • Your build server can enforce that your unit tests pass and generate reports
- 19. Unit Testing with tSQLt / SQL Test Demo Commit Changes via SQL Source Control CI Server Detects Changes Kicks off Build Process Build Task executes Unit Test procedures Tests can pass or fail the build
- 20. Migrations
- 21. Springer’s Final Thought • Saves developers time (yay!) – No more maintaining update scripts – Don’t need to deploy scripts when QA needs changes • Identifies holes in your deployment process – Are developers or DBAs making changes directly to production? – Are indexes/constraints missing from your Dev/QA/Prod environments? • Creates a definitive database build that can be easily deployed and redeployed
- 22. Considerations • Environments must be pristine • Process must be changed and understood • Production change scripts should be scrutinized (especially for the first few releases) • Migration Scripts can be used to massage data • DB Replication requires more work for promotions
- 23. Appendix 1: Other CI Servers • • • • Jenkins (open source) Team City (JetBrains) Bamboo (Atlassian) CruiseControl / CruiseControl.NET (open source)
- 24. Appendix 2: Other Source Control Systems • • • • • • Subversion (SVN) Git Team Foundation Server Mercurial (Hg) AccuRev Surround SCM
- 25. Links / Contact Info • • • [email protected] @ernestedcode https://github.com/CF9/Databases.RGDemo • Practice Fusion is Hiring (email me) http://practicefusion.com/careers/ facebook.com/practicefusion @practicefusion • •