7.3 Kanban, XP, and SAFe
Key Takeaways
- Kanban prescribes no roles, iterations, or ceremonies; its defining feature is explicit Work-in-Progress (WIP) limits that pull work and expose bottlenecks
- Little's Law links Kanban metrics: average Cycle Time = average WIP / average Throughput
- Extreme Programming (XP) emphasizes engineering practices — pair programming, test-driven development, continuous integration, refactoring — guided by five values
- SAFe scales agile across many teams using the Agile Release Train (50-125 people) and a Program Increment of typically 8-12 weeks set in PI Planning
- Lean's seven principles (Poppendiecks) — eliminate waste, amplify learning, decide late, deliver fast, empower the team, build integrity in, optimize the whole — underpin all agile
Kanban
Kanban (Japanese for "signboard") is a lean, flow-based method for managing knowledge work. Unlike Scrum, it prescribes no roles, no fixed iterations, and no ceremonies — you overlay it on your existing process and improve evolutionarily.
The core practices
| Practice | What it means |
|---|---|
| Visualize the workflow | A Kanban board shows every item and its stage |
| Limit Work in Progress (WIP) | Cap items per column to stop overloading |
| Manage flow | Optimize the smooth movement of work |
| Make policies explicit | Document the rules (e.g., "Done = peer reviewed") |
| Implement feedback loops | Cadenced reviews to inspect and adapt |
| Improve collaboratively, evolve experimentally | Continuous, incremental change |
WIP limits — the defining feature
A column with a WIP limit of 3 accepts no new item until one moves forward. This is a pull system: capacity, not a manager, decides when work starts. WIP limits expose bottlenecks (work piles up before the constrained stage) and shorten cycle time.
Flow metrics and Little's Law
| Metric | Definition |
|---|---|
| Lead Time | Customer request to delivery |
| Cycle Time | Work start to work finish |
| Throughput | Items completed per period |
| Cumulative Flow Diagram (CFD) | Items in each stage over time |
Little's Law ties them together: average Cycle Time = average WIP / average Throughput. Cutting WIP (with everything else equal) reduces cycle time — a common CAPM/PMI-ACP fact. Kanban is the natural choice for support, maintenance, and operations work with a continuous, unpredictable intake.
Extreme Programming (XP)
Extreme Programming (XP), created by Kent Beck, pushes technical excellence to the extreme for software teams. Its five values are Communication, Simplicity, Feedback, Courage, and Respect.
| XP Practice | Description |
|---|---|
| Pair Programming | Two developers, one workstation, continuous review |
| Test-Driven Development (TDD) | Write the failing test before the code |
| Continuous Integration (CI) | Integrate and test many times per day |
| Refactoring | Improve structure without changing behavior |
| Small/Frequent Releases | Short cycles for rapid feedback |
| Collective Code Ownership | Anyone can improve any code |
| Coding Standards | Shared conventions |
| Simple Design | The simplest design that works |
| On-Site Customer | A customer available to the team |
| Sustainable Pace | Roughly 40-hour weeks; no death marches |
CAPM links XP to engineering quality: if a question describes failing builds, defect-prone code, or no automated tests, the XP practices (CI, TDD, pairing, refactoring) are the answer — not a new Scrum role.
SAFe (Scaled Agile Framework)
SAFe scales agile to dozens or hundreds of people across many teams building one large solution.
| Level | Focus | Key construct |
|---|---|---|
| Team | Individual agile teams | Scrum or Kanban teams |
| Program (ART) | Teams aligned to a value stream | Agile Release Train (ART) |
| Large Solution | Multiple ARTs | Solution Train |
| Portfolio | Strategy and funding | Lean Portfolio Management |
Key SAFe terms
| Concept | Definition |
|---|---|
| Agile Release Train (ART) | A team-of-teams, 50–125 people, that plans and delivers together |
| Program Increment (PI) | A planning interval, typically 8–12 weeks, of several iterations |
| PI Planning | A cadenced event where the whole ART aligns on PI objectives |
| Architectural Runway | Existing tech enabling near-term features |
| Innovation & Planning (IP) Iteration | Buffer iteration for innovation and slack |
| WSJF | Weighted Shortest Job First — SAFe's prioritization formula |
Use SAFe when many teams must coordinate; it is overkill for a single small team (a frequent distractor).
Lean Principles
Lean thinking (Mary and Tom Poppendieck, Lean Software Development) underpins agile with seven principles:
- Eliminate waste — remove anything not adding value.
- Amplify learning — short feedback loops.
- Decide as late as possible — keep options open to the last responsible moment.
- Deliver as fast as possible — short cycle times.
- Empower the team — give autonomy.
- Build integrity in — quality is everyone's job.
- Optimize the whole — improve the full value stream, not silos.
Choosing a Framework
| Feature | Scrum | Kanban | XP | SAFe |
|---|---|---|---|---|
| Cadence | Fixed Sprints | Continuous flow | Short releases | PI (8–12 wks) |
| Roles | PO, SM, Developers | None prescribed | Coach, Customer, Developers | Many |
| Defining feature | Time-boxed iterations | WIP limits | Engineering practices | ART + PI Planning |
| Best for | Most team projects | Support/maintenance flow | Technical quality | Large enterprises |
Scrumban and Combining Frameworks
Real teams often blend frameworks, and the exam expects you to recognize the hybrids. Scrumban keeps Scrum's roles and cadence but adds Kanban's WIP limits and pull-based flow — popular for maintenance teams transitioning off Scrum or Scrum teams drowning in unplanned work. Many teams also run Scrum with XP engineering practices: Scrum supplies the management wrapper (roles, events, backlog) while XP supplies the technical discipline (TDD, continuous integration, pairing). These are not contradictions; Scrum is intentionally silent on engineering technique, so layering XP fills the gap.
When a question pairs a delivery cadence with a technical-quality problem, the strongest answer often combines the two rather than choosing only one.
Choosing Among the Frameworks — Worked Scenarios
Consider three quick scenarios. A help-desk team handles a steady stream of unpredictable tickets and wants to cap simultaneous work and surface bottlenecks. The pull system, WIP limits, and lack of fixed iterations make Kanban the fit. A 90-person product organization with eight teams must release one coordinated solution every quarter. The need to align many teams on a cadenced planning interval points to SAFe, with its Agile Release Train and Program Increment. A single eight-person team building a new product with evolving requirements wants regular feedback and a simple, well-known structure. That is textbook Scrum.
Matching the scale and the dominant problem — flow, scaling, technical quality, or iterative delivery — to the framework's defining feature is exactly how CAPM frames these items, so anchor your choice to that defining feature rather than to surface keywords.
A maintenance team has continuous, unpredictable ticket intake and wants to expose bottlenecks and shorten cycle time without imposing fixed iterations. Which framework fits BEST?
By Little's Law, if a Kanban team reduces its average Work in Progress while throughput stays constant, what happens to cycle time?
In SAFe, which statement about the Agile Release Train (ART) and Program Increment (PI) is correct?
A software team suffers frequent broken builds and defect-prone code with no automated tests. Which set of practices most directly addresses this?