C5: Load Testing

Objective

Do now: Validate that the revised FreshConnect platform can handle realistic peak load.
Input: Deployed endpoint from C3/C4, or a documented fallback plan if deployment is blocked.
Output: agent-output/freshconnect/05-load-test-results.md.
Required to move on: Test scenario, targets, observed results, and a recommendation.
Decisions now: Which endpoint matters most, how to ramp to 500 users, what counts as pass or fail, and what to change next if results miss target.
Next: C6 uses the results to create audience-specific operational documents.

This challenge is about decision-quality, not just running a tool. The report should tell the next reader what you tested, what happened, and what it means for launch risk.

The Business Challenge

Nordic Fresh Foods expects peak-season demand of about 500 concurrent users during the summer and holiday rush. Before go-live, the team needs evidence that the platform can stay responsive enough to protect conversion and customer trust.

Metric	Target	Why it matters
Concurrent users	500	Represents expected peak demand
P95 response time	`<= 2 seconds`	Slow pages drive abandonment
Error rate	`< 1%`	Frequent failures undermine trust
Sustained duration	5 minutes	Short spikes do not prove steady performance

Your Tasks

Pick the endpoint or user journey that best represents real customer traffic.
Run a load test with k6 as the default path; use Azure Load Testing only if your team already has time and confidence.
Compare the results to the target thresholds and decide whether the bottleneck is acceptable, fixable now, or a recommendation for phase 2.
Save a concise report at agent-output/freshconnect/05-load-test-results.md.

Key Decisions

Which endpoint gives the most honest picture of real user experience: homepage, API, or another critical flow?
If results fail, is the likely bottleneck compute, data, networking, or poor test design?
What scale or tuning change would you recommend first, and what is the likely cost impact?
If you have no deployed endpoint, what is the minimum credible fallback test plan you can still document for later execution?

Deliverables

agent-output/freshconnect/05-load-test-results.md
Test configuration: endpoint, stages or concurrency shape, duration, and thresholds.
Results summary with pass or fail against the stated targets.
Short interpretation of the main bottleneck, headroom, or risk.
Recommendation for what to do next if performance misses target.

Success Criteria

Focus	What good looks like	Evidence
Realistic test design	The scenario reflects a meaningful workload instead of a trivial ping	Endpoint, thresholds, and duration are documented clearly
Results capture	The team records enough data to support a decision	Summary includes P95, error rate, concurrency, and pass/fail status
Interpretation	The report explains what the numbers mean	Likely bottleneck or headroom is described in plain language
Recommendation	The next action is clear	Report states keep, tune, scale, or defer with rationale

Tips / Hints

Quick test pattern

k6 is the fastest default path. A minimal run looks like this:

k6 run load-test.js

Keep the report focused on signal, not raw terminal output. Use Hints & Tips for deeper test ideas and Troubleshooting if the endpoint is unavailable.

Watch Out

A health endpoint alone is usually too shallow to represent customer experience.
A failing test without interpretation is not useful evidence.
Do not paste pages of raw output into the report without summarizing it.
If no deployment exists, clearly label the fallback as an intended test plan rather than a completed execution.

Artifact Handoff

Item	Value
Input from	Deployed endpoint and platform design from C3/C4, or documented fallback plan
Your output	`agent-output/freshconnect/05-load-test-results.md`
Next challenge uses	C6 uses the results and recommendations to produce audience-specific documentation

Next Step

Challenge 6 turns your technical evidence into operational documents. A good load-test report gives the documentation step concrete numbers, risks, and actions to explain.