20.1 Lifecycle Mental Model
1. The lifecycle
Treat every Datasphere capability as a controlled lifecycle:
Design -> Build -> Validate -> Package -> Transport -> Deploy -> Operate -> Measure -> Change/Retire
The lifecycle has four dimensions:
| Dimension | Question |
|---|---|
| Metadata | What objects and dependencies change? |
| Runtime | What must be configured or executed? |
| Data | What data must exist and be reconciled? |
| Control | Who approves, transports and operates it? |
2. Design-time versus runtime
A transport package primarily carries definitions. Runtime readiness can still depend on target connections, credentials, permissions, schedules, mappings and source-system availability.
Therefore use a release checklist rather than assuming import equals production readiness.
3. Immutable release thinking
Prefer a versioned package and controlled promotion over editing production objects directly. Emergency changes should be exceptional, documented and reconciled back into the source landscape.
4. Definition of done
A release is done only when:
- objects are deployed successfully;
- dependencies resolve;
- security is correct;
- required runtime settings exist;
- data freshness is within SLO;
- reconciliation passes;
- monitoring is active;
- business validation is complete;
- rollback or recovery is understood.