C4: DR Curveball
Objective
Section titled “Objective”- Do now: Respond to the DR curveball with an ADR, updated design, and revised delivery path.
- Input: C3 templates, deployment outcome, implementation plan, and architecture diagram.
- Output:
agent-output/freshconnect/03-des-adr-NNNN-ha-dr-strategy.md, updated IaC or paper design, updated architecture diagram, and revised cost view. - Required to move on: Revise affected artifacts, rerun required reviews, approve the new path, parameterize the design, and document whether you deployed it or designed it on paper.
- Decisions now: Single-region HA vs multi-region DR vs active-active, what must replicate, how failover works, and what fits inside the extra budget.
- Next: C5 validates the revised platform or documents the intended test target if you stayed on paper.
This challenge tests whether your team can adapt without losing the architectural thread. The right answer is the option you can justify and deliver credibly under time pressure.
The Business Challenge
Section titled “The Business Challenge”Midway through the workshop, Nordic Fresh Foods signs a major Danish contract and the
board raises the budget ceiling to about €700 per month. The platform now needs a
secondary region in germanywestcentral, an RTO of 1 hour, and an RPO of 15 minutes.
You must recommend a resilient path quickly, then show how it changes the design.
Your Tasks
Section titled “Your Tasks”- Record the changed business requirement and identify which approved artifacts it invalidates.
- Return to
02-Requirementsand03-Architectas needed. Update requirements, recovery targets, architecture, cost, and SKU evidence. - Select
10-Challengerfor the required revised requirements, architecture, and cost reviews. Resolve blocking findings and approve the updated design. - Reconcile governance, revise the IaC plan with
05-IaC Planner, complete the plan review, and approve the revised code-generation contract. - Select
04-Designto write the ADR asagent-output/freshconnect/03-des-adr-NNNN-ha-dr-strategy.md(next ADR number) with context, decision, consequences, and rejected alternatives. - Update the selected IaC track or document the parameterized paper design. Validate it before any deployment request.
- Explicitly authorize the revised deployment scope, or state that the result remains a paper exercise. Update the diagram and cost view in both cases.
| Challenge 3 outcome | What you do now |
|---|---|
| Deployment succeeded | Revise affected upstream artifacts and approvals, then extend the deployed platform and capture evidence |
| Partial deployment | Revise and validate what works, document gaps, and show the approved intended end state |
| Deployment failed | Complete the reviewed paper path without claiming deployment evidence |
Key Decisions
Section titled “Key Decisions”- Which services truly need redundancy to meet the new RTO and RPO, and which can be recovered more simply?
- Is active-passive good enough for the business, or are you paying for active-active complexity you cannot defend?
- Which failover steps are automatic, which are manual, and how will the team know the difference during an incident?
- What proof will convince a stakeholder that this DR plan is credible within the new budget?
Deliverables
Section titled “Deliverables”agent-output/freshconnect/03-des-adr-NNNN-ha-dr-strategy.md- Updated IaC with HA/DR parameters, or a written parameterized design if you are on the paper path.
- Updated architecture diagram showing regions, replication paths, and failover flow.
- Revised cost estimate or cost assumptions for the chosen approach.
- Deployment evidence if you applied the DR change.
Success Criteria
Section titled “Success Criteria”| Focus | What good looks like | Evidence |
|---|---|---|
| Decision quality | The team chooses an HA/DR path with clear business reasoning | ADR states the trigger, chosen option, trade-offs, and rejected alternatives |
| DR design clarity | The resilience design is concrete instead of hand-wavy | Parameters, regional changes, and failover behavior are explicit |
| Delivery path | The team is honest about what was deployed versus designed | Deployment evidence exists, or the paper path is documented cleanly |
| Architecture communication | Others can understand the updated design quickly | Updated diagram and ADR tell the same before/after story |
Tips / Hints
Section titled “Tips / Hints”Useful reference points for the curveball
Use Hints & Tips for DR design prompts, Quick Reference Card for the post-curveball budget guardrail, and Quick Reference Card if your team is documenting the fallback path.
ADR skeleton:
Context -> what changed and why it mattersDecision -> which HA/DR option you choseConsequences -> cost, complexity, risks, operational impactAlternatives -> what you rejected and whyWatch Out
Section titled “Watch Out”- A paper exercise can still score well, but only if you separate design intent from actual deployment evidence.
- The paper path can still prove your C4 design thinking, but it does not replace missing C3 deployment evidence. Be explicit about what you designed versus what ran.
- Do not patch downstream code while approved requirements, architecture, or plans still describe the old recovery targets.
- Do not pick active-active unless you can explain consistency, failover, and cost.
- Do not update the diagram without updating the ADR, or vice versa.
- Keep the €700 ceiling visible; the extra budget is not a license to duplicate everything blindly.
Artifact Handoff
Section titled “Artifact Handoff”| Item | Value |
|---|---|
| Input from | C3 templates, deployment outcome, implementation plan, architecture diagram |
| Your output | agent-output/freshconnect/03-des-adr-NNNN-ha-dr-strategy.md, updated IaC or paper design, updated diagram |
| Next challenge uses | C5 validates the revised platform if you have an endpoint, or uses your documented target state to define the intended test plan |
Next Step
Section titled “Next Step”Challenge 5 treats this DR-aware platform as the system under test. If you deployed the change, you will validate it under load; if you stayed on paper, you will document the test plan and expected thresholds against the design you proposed in C4.