⚡ Performance Tester interview prep

// WHO THIS IS FOR

For engineers interviewing for roles focused on non-functional testing: designing and executing load, stress, spike, and soak tests, defining and validating NFRs, and diagnosing system bottlenecks from test results. Covers mid-to-senior scopes where the interviewer expects you to own the full performance testing cycle — not just run a tool, but interpret results and drive remediation.

// SKILLS INTERVIEWERS EXPECT

Green = most frequently tested

// TYPICAL INTERVIEW ROUNDS

1. 1

   Performance fundamentals

   30–45 min questions on the taxonomy of performance tests (load, stress, spike, soak, volume, capacity), how NFRs are defined and documented, and the difference between SLAs, SLOs, and SLIs. Interviewers also probe why average response times are misleading and when to use percentiles, what think time is and why it matters for simulation realism, and how you baseline a system before a load test.

2. 2

   Test design and scripting

   Given a system description and an NFR (e.g. 1,000 concurrent users, p95 < 2 s), design a test plan: user journeys to simulate, ramp-up shape, data set requirements, and correlation points (dynamic tokens, session IDs). You may be asked to write or review a JMeter test plan or a k6 script live. Interviewers look for realistic load shape design, not just 'run threads until it breaks'.

3. 3

   Results analysis and diagnosis

   Given a Grafana dashboard or a JMeter results file showing degrading response times, error spikes, and throughput plateaus, identify the likely bottleneck (CPU saturation, connection pool exhaustion, GC pressure, slow database queries, external dependency latency). Produce a structured findings report — what you observed, what it indicates, how you would confirm the root cause, and what remediation options exist.

4. 4

   Behavioural

   Situational questions — a load test reveals a critical bottleneck two days before a release, stakeholders want to declare the system 'good enough' when p99 is 8 s, or a test environment is too small to simulate production traffic. Also covers how you communicate performance findings to engineers and product stakeholders, and how you advocate for fixing NFR gaps when performance is deprioritised.

// TOPICS TO STUDY

• Performance test types: load, stress, spike, soak, volume, capacity — when to use each
• NFRs: how to define, document, and validate response time and throughput SLAs
• Percentile metrics: p50, p75, p95, p99 — why averages mask tail latency problems
• JMeter: thread groups, ramp-up periods, HTTP Sampler, CSV Data Set Config, Assertions, Listeners
• k6: VU lifecycle (init, setup, default, teardown), stages, thresholds, checks, custom metrics
• Think time and pacing: simulating realistic user cadence to avoid artificial thundering herds
• Correlation: extracting dynamic values (CSRF tokens, session IDs) with regex/JSON extractors
• Bottleneck identification: CPU/memory saturation, DB slow queries, connection pool limits, GC pauses
• Monitoring stacks: Grafana + InfluxDB for real-time dashboards during test execution

// READINESS SELF-CHECK