25.14 Deployment Pipelines
Start with a problem
Raja publishes untested changes straight to the workspace managers use.
Explain simply
Deployment pipelines move content through Development → Test → Production workspaces with rules for different data sources.
Show the data
| Stage | Workspace | Who |
|---|---|---|
| Development | Sales [Dev] | Raja |
| Test | Sales [Test] | Shraddha |
| Production | Sales | Store managers |
Fictional sample data. Practice file: visuals_orders_apr_may_2026.csv (source: src/practice/).
Demonstrate
- Deployment pipelines › Create pipeline.
- Assign Dev, Test and Prod workspaces.
- Deploy Dev → Test; Shraddha tests.
- Deploy Test → Prod.
Show the result
Your result should look like this (fictional sample):
| Stage | Card |
|---|---|
| Dev | ₹2,100 (test data) |
| Prod | Real data |
Explain what happened
Deployment copies items forward; rules switch sources per stage.
Explore further
- Needs Premium/Fabric capacity or PPU (check current licensing).
Detailed reference for this lesson:
Dev, test, production. Tine workspace. Pipeline tyanna jodto. Ha capacity feature aahe.
A deployment pipeline promotes content from Development to Test to Production. It needs a capacity that includes pipelines (Premium Per User, or a Fabric capacity; the licence names change). Section 25.3 already said to keep separate workspaces. The pipeline is the button that copies the reports and models forward instead of you downloading a PBIX and re-uploading it.
Steps
- Create three workspaces:
MH QC Dev,MH QC Test,MH QC Prod. - Deployment pipelines › New pipeline. Assign the three workspaces to the three stages.
- Build in Dev. Deploy to Test. Compare. Then deploy to Prod.
- Use rules on the dataset parameters (server name, section 10.4) so Test does not point at the production database after a deploy.
Deploy copies metadata. It does not excuse an untested measure. RLS roles deploy; the members of those roles in the Service do not always follow the way you expect. Re-check Security on the production semantic model after the first deploy.
Worked example. Dev uses the fictional sample. Test uses a copy of that sample so Shraddha can click through Kolhapur. Prod will later use the real server, which is outside this course. A parameter rule on ServerName is what stops Test from reading Prod after you deploy. Check the parameter in the Test model settings before you invite Rani.
Ravindra Bagale's Tip
Many students deploy Dev straight to Prod in the first week, with no rules, and the production refresh hits the wrong server. Do not skip the Test stage. Check the parameter rule. Pay attention!
Ravindra Bagale's Tip – मराठी
बरेच students पहिल्याच आठवड्यात कोणतेही rules न ठेवता Dev सरळ Prod ला deploy करतात, आणि production refresh चुकीच्या server वर जातो. Test stage skip करू नका. Parameter rule बघा. लक्षात ठेवा!
Ravindra Bagale's Tip – हिंदी
बहुत से students पहले ही हफ़्ते बिना किसी rule के Dev सीधे Prod पर deploy कर देते हैं, और production refresh ग़लत server पर जाता है. Test stage skip मत करो. Parameter rule देखो. ध्यान रखो!
Fix common mistakes
| Wrong result | Reason | Correction |
|---|---|---|
| Prod shows test data | No deployment rule | Add data source rule |
| Users see unfinished pages | Editing in Prod | Edit in Dev |
| Pipeline missing | Licence | Check capacity |
Practise independently
Practice task
Where should Raja test a new measure?
Hints
- Not production.
Solution
In Dev, then Test before Prod.
Think like a professional
Dev/Test/Prod discipline is expected in enterprise roles.
Revision card