Skip to content
Tutorials Center
My learning

Topics

Lesson 9 of 50 ยท The SDLC and Process Models

Incremental, RAD, Iterative Waterfall, DSDM and Crystal

Compare the incremental model, rapid application development (RAD), iterative waterfall, DSDM and Crystal: phases, strengths, weaknesses and when to pick each one.

Between strict waterfall and full agile sits a family of models that all share one idea: do not build everything in one pass. Deliver in pieces, learn from each piece and adjust. This lesson puts five of them side by side: the incremental model, the iterative waterfall, rapid application development (RAD), DSDM and the Crystal family.

The examples are small plain-JavaScript calculators with invented numbers. They illustrate the ideas; no real project tool is behind them.

The incremental model

In the incremental model you split the product into increments, each a usable slice of the final system. Every increment goes through its own mini-cycle of analysis, design, coding and testing, and each one is added on top of the previous release.

  • Increment 1 is the core product: the features without which the product is pointless.
  • Later increments add features, ordered by value and risk.
  • The customer can use increment 1 while increment 2 is still being built.

A variant is staged delivery: the requirements and architecture are fixed up front (as in waterfall), but construction and release happen in stages. You get early value without giving up upfront design.

Strengths: early working software, easier testing of small pieces, early feedback, lower risk of a total failure, and the team can start small and grow.

Weaknesses: the architecture must be planned well enough up front that later increments plug in cleanly; integration problems appear if interfaces were poorly defined; total cost can be higher than a single build because of repeated integration and regression testing.

The iterative waterfall model

Classic waterfall has no way back. The iterative waterfall adds feedback loops: when a defect or a misunderstanding is found in a later phase, the team goes back to the phase that caused it, fixes the artifact and moves on again. Coding bugs go back to design, design flaws go back to requirements.

It is still one big plan, so the loops are for corrections, not for new learning. Rework is cheaper than in pure waterfall because it is allowed, but late discoveries still cost more than early ones. Its main weakness is unchanged: the customer sees the product only at the end.

Rapid application development (RAD)

RAD aims to deliver a working system very fast, often in 60 to 90 days, by trading up-front planning for prototyping, reuse and heavy user involvement. Its typical phases:

  1. Business modeling: map the information flows between business functions.
  2. Data modeling: define the data objects and how they relate.
  3. Process modeling: describe how data is created, read, updated and deleted.
  4. Application generation: build with components, code generators and visual tools.
  5. Testing and turnover: component tests are reused, so mainly the interfaces and the whole are tested; the finished system is handed over (the "cutover").

Large systems are split into modules that separate teams build in parallel; each module is built in a time box, then everything is integrated.

Strengths: fast delivery, high customer satisfaction through constant involvement, adaptable requirements, reuse of components, lower risk of building the wrong thing.

Weaknesses: it needs skilled, available people and users who really commit time; it fits only systems that can be modularized; it is a poor choice for high technical risk or very large systems; a rushed result can be weak on performance and long-term quality.

Situation RAD fits A plan-driven model fits
Time pressure Short, hard deadline Schedule has slack
Requirements Understood in outline, details flexible Fixed, regulated, documented up front
Users Available and engaged Distant or unavailable
Technology Well known, low risk New, complex or safety critical
Size Small to medium Large, tightly coupled

DSDM: the Dynamic Systems Development Method

DSDM is an agile framework that grew out of RAD. Its distinctive rule: fix time and cost, vary the scope. Requirements are prioritized with MoSCoW and the team protects the "must" items while "should" and "could" items flex.

Letter Meaning Rule of thumb
M Must have Without it the release is a failure
S Should have Important, but a workaround exists
C Could have Nice to have, first to drop
W Won't have (this time) Agreed out of scope

DSDM adds roles (business sponsor, ambassador user, project manager, solution developer and tester), time-boxed iterations, frequent delivery and a focus on the business need. Teams commonly keep the "must" share at or below about 60 percent of effort so that the time box has a safety buffer.

JS MoSCoW time box planner

Press Run (or Ctrl + Enter) to see the console output here.

No output.

Crystal methodologies

Crystal is a family of lightweight methods by Alistair Cockburn. The central claim is that no single process fits every project, so the method is chosen by team size and by criticality (what damage a failure can do). The families are named by color, from clear (small teams) through yellow, orange and red up to larger, higher-risk projects. Crystal Clear, for teams of about 6 to 8 co-located people on low-risk work, is the best known.

Crystal stresses people over process: frequent delivery, close communication, reflective improvement workshops, and "osmotic" communication, which means information flows by people sitting near each other. Its weaknesses are a smaller community and less detailed guidance than Scrum, and it depends heavily on a good, co-located team.

Comparing the models

JS Compare process models

Press Run (or Ctrl + Enter) to see the console output here.

No output.

When deciding, ask three questions: how stable are the requirements, how much risk is in the technology, and how available are the users? Stable, regulated work leans plan-driven; uncertain, user-driven work leans incremental, RAD or agile. Choosing a process goes deeper.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What does the incremental model deliver first?

    Each increment is a working slice. The first one holds the core features.
  2. 2. What is the difference between waterfall and iterative waterfall?

    The loops let the team return to the phase that caused a problem, but it is still one big plan.
  3. 3. In DSDM what is fixed and what varies?

    DSDM protects the time box and budget and lets the lower-priority scope flex, guided by MoSCoW.
  4. 4. Which project is a poor fit for RAD?

    RAD needs modularity, skilled people and low technical risk. Safety-critical systems need planned, verified design.

Summary

  • The incremental model delivers working slices; staged delivery fixes the design first and releases in stages.
  • Iterative waterfall adds feedback loops for corrections but keeps one big plan.
  • RAD uses prototyping, reuse, parallel module teams and engaged users to deliver in a very short time.
  • DSDM fixes time and cost and prioritizes scope with MoSCoW.
  • Crystal picks a method by team size and criticality and puts people first.

Next: Choosing a process.