1.3 Strategies

Different approaches structure how a solution is designed and built. Choose an approach based on complexity, team size, reuse needs, and available components.

• Top-down:
  Start with the whole problem, break it into major parts, then into smaller tasks until each can be implemented. Used when the overall goal is clear and you need a coherent end-to-end design.
• Bottom-up:
  Start by creating or selecting small, reusable components, then combine them into larger subsystems and finally the whole solution. Used when you have libraries/services to build upon or when rapid prototyping of parts is possible.
• Modularisation:
  Organise the solution into independent, well-defined modules with clear interfaces. Used to enable parallel work, easier maintenance, testing, and reuse.

• Top-down – Benefits:
  Clear roadmap, aligns teams to a shared goal, helps manage complexity.
  Drawbacks:
  Early design mistakes can cascade; may overlook low-level constraints until late.
• Bottom-up – Benefits:
  Encourages reuse, quick wins from working components, realistic constraints discovered early.
  Drawbacks:
  Risk of misalignment with the big picture; integration can be hard.
• Modularisation – Benefits:
  Easier testing, maintenance, and parallel development; clear ownership.
  Drawbacks:
  Requires good interface design and governance; too many modules can add overhead.

RCA is a structured method to identify the underlying cause(s) of an issue so you can prevent recurrence, not just fix symptoms.

After incidents, recurring issues, safety/security events, quality problems, or failed changes where long-term prevention is required.

• Five Whys:
  Ask “why?” repeatedly (typically five times) to move from symptom to underlying cause.
• Failure Mode and Effects Analysis (FMEA):
  Systematically list potential failure modes, their effects, causes, and controls; prioritise using severity, occurrence, and detection.
• Event Tree Analysis (ETA):
  Start from an initiating event and map forward through branches of system responses to evaluate possible outcomes and their likelihood.

Implement corrective and preventive actions, log outcomes, document residual risks, communicate changes, and escalate to the appropriate manager where policy, budget, or cross-team changes are required. Close the record when actions are verified.

1. Define the problem:
   Clarify scope, impact, constraints, and success criteria.
2. Gather information:
   Collect logs, data, user reports, environment details, and recent changes.
3. Analyse the information:
   Form hypotheses, compare expected vs actual behaviour, identify likely causes.
4. Make a plan of action:
   Prioritise safe, reversible steps aligned with risk and impact.
5. Implement a solution:
   Execute the plan, communicate status, and track results.
6. Review the solution:
   Verify outcomes, document lessons learned, and prevent recurrence.

is a single unplanned event that disrupts service operations and negatively impacts service quality (e.g., outage, degraded performance, security alert).

is the underlying cause of one or more incidents (e.g., a software defect, misconfiguration, or capacity shortfall) that must be identified and resolved to prevent recurrence.

• Detection:
  Report, record, categorise, and prioritise the incident; assess impact and urgency.
• Response:
  Assign an owner, diagnose, resolve and restore service, apply workarounds as needed, and record the resolution.
• Intelligence:
  Capture lessons learned, identify root cause or link to a problem record, and share insights with stakeholders.

Problems, incidents, and strategies are linked: recurring incidents point to deeper problems that require structured strategies (e.g., RCA, modular fixes) rather than ad-hoc patches. Selecting an approach depends on impact, risk, time, available components, and required long-term reliability.

severity and frequency of incidents, clarity of the end goal, availability of reusable modules, constraints (security, compliance), and the need for quick restoration vs permanent fix.