Learning Objectives
By the end of this lesson you will be able to:
- Describe how PDM relates to waterfall, agile, and blended delivery approaches
- Explain why pure waterfall and pure agile both present challenges for data migration
- Describe the blended approach and how PDM’s release management supports it
The Methodological Question
When a project asks “are we doing waterfall or agile?”, they are asking about how they will organise and deliver work. This is a governance decision captured in the MSG.
PDM is compatible with all three approaches. But each has implications for how PDM processes are organised, and each has specific risks in the data migration context.
Waterfall
In a waterfall approach, project phases are sequential: analysis completes before design begins, design completes before build begins. Each phase has formal sign-off gates.
Works well with PDM because:
- The natural progression of PDM (MSG → LA → GAM → DQR → MDE → LD) has a broadly sequential logic
- Business sign-off gates align with the formal governance culture of many waterfall projects
- The LDS list and mapping documents are naturally phased deliverables
Challenges for data migration:
- Data quality issues discovered during build or test often require rework of mapping documents - which is hard to accommodate in a strict waterfall gating model
- The DQR process is inherently iterative: issues are raised continuously throughout the project, not just in a defined analysis phase
- New LDS are discovered throughout the project, challenging a fixed-scope waterfall structure
The key PDM mitigation: treat the overall project as broadly sequential, but design specific phases (particularly DQR and LA) as iterative activities that continue across phase boundaries.
Agile
In an agile approach, work is delivered in short sprints with continuous re-prioritisation. The backlog evolves; scope is not fixed.
Works well with PDM because:
- DQR items resemble a product backlog: continuously discovered, prioritised, and worked to completion
- Fortnightly release cycles fit naturally into a sprint cadence
- The emphasis on working software (data that meets quality thresholds) aligns with agile’s delivery focus
Challenges for data migration:
- Data migration has a non-negotiable endpoint: go-live. The “continuous delivery” model of agile does not map onto a one-time migration event
- The business engagement workstream requires committed time from Data Owners and BDEs - which is harder to sustain in an agile model where scope and timing are fluid
- Legacy decommissioning requires formal sign-off, which conflicts with the informality of most agile governance
Pure agile is not recommended for data migration programmes. The one-time, irreversible nature of a migration requires more formal governance than agile typically provides.
Blended Approaches
Most large data migration projects use a blended approach: the overall programme is structured in phases (broadly waterfall), while the analytical and build work within each phase uses iterative, sprint-like cycles.
PDM is designed for this. The release management sub-module provides the iterative delivery cadence (fortnightly releases, push-planned by governance). The DQR process provides the continuous backlog management. The BTR interview process provides the formal business engagement. The overall programme plan provides the sequential phasing.
Implications for PDM Implementation
When PDM is embedded in a supplier-led programme that uses agile:
- The DQR Board becomes the bridge between PDM governance and the supplier’s sprint planning
- Release content (PDM terminology) maps to sprint scope (agile terminology)
- BTR sign-offs become formal user stories or acceptance criteria in the supplier’s backlog
- The Migration Analyst becomes a liaison between the business engagement workstream and the delivery team
When embedded in a waterfall programme:
- PDM’s iterative elements (DQR, LA discovery) must be explicitly accommodated within or alongside the phase gates
- Change control processes must account for the ongoing discovery of new LDS and DQR items
Key Takeaways
- PDM is compatible with waterfall, agile, and blended approaches
- Pure waterfall struggles with PDM’s inherently iterative elements (DQR, LDS discovery)
- Pure agile struggles with the formal governance that decommissioning and go-live require
- The blended approach - sequential phases with iterative cycles within - is the recommended model
- Release Management provides the iterative delivery cadence that bridges waterfall and agile
Book Reference
Practical Data Migration by Johny Morris (BCS, The Chartered Institute for IT):
- Chapter 15 - Agile, Waterfall and Blended