Practice: Analyze a Load Test and Find the Bottleneck

Time to put everything together. In this exercise, you will run a load test against the Shopping Portal, analyze the results, identify bottlenecks, and write a mini performance report. This is the closest thing to what you will do on a real project.

Exercise 1: Run and Analyze a Load Test

Exercise 2: Analyze a Simulated Scenario

Even without running a test, you can practice analysis skills. Study the following simulated data and answer the questions below.

Test Configuration:
- 300 concurrent users, ramp-up over 3 minutes, hold for 10 minutes
- 4 endpoints: /login, /dashboard, /api/transactions, /api/transfer

Aggregate Results:
────────────────────────────────────────────────────────────
Endpoint             Samples  Avg(ms)  p50   p95    p99    Err%
/login                12,000    180    150    420    800   0.1%
/dashboard            30,000    220    180    550   1,200   0.2%
/api/transactions     25,000    450    300  1,800   4,500   1.5%
/api/transfer          8,000  2,100    800  8,200  15,000   6.8%
────────────────────────────────────────────────────────────

Server Metrics During Peak Load:
- App Server CPU: 55%
- App Server Memory: 72% (stable)
- DB Server CPU: 94%
- DB Active Connections: 19 of 20 (connection pool almost full)
- DB Slow Query Log: SELECT * FROM accounts WHERE user_id = ?
  AND status = 'active' -- avg 1,200ms, called 33,000 times

Error Details:
- 95% of errors are HTTP 504 (Gateway Timeout) on /api/transfer
- 5% of errors are HTTP 500 on /api/transactions

Analysis Questions

1. Which endpoint is the primary bottleneck? How can you tell?
2. Is the bottleneck in the application server or the database? What evidence supports your answer?
3. What is the likely root cause of the /api/transfer slowness?
4. Why are most errors 504 Gateway Timeout rather than 500 Internal Server Error?
5. What specific fix would you recommend first? Why that one before others?
6. Bonus: Using Little's Law, is the system behaving as expected at 300 users? (Hint: calculate expected throughput)

Exercise 3: Write a Findings Table

Using the simulated data above, create a findings table with the following SLA targets: Login p95 < 500ms, Dashboard p95 < 1 second, Transactions p95 < 2 seconds, Transfer p95 < 3 seconds, Overall error rate < 1%, System must support 300 concurrent users. Mark each as PASS or FAIL and write a one-paragraph executive summary with a go/no-go recommendation.