MeasureUp Test Performance Analysis 2 — Questions and Answers
Question 1: Which metric directly measures the percentage of time a CPU is actively executing instructions rather than waiting for I/O?
- CPU utilization (Correct answer)
- Throughput
- Latency
- Queue depth
Correct answer: CPU utilization
CPU utilization measures the fraction of time the processor spends on non-idle work.
Question 2: In a performance baseline, what is the primary purpose of collecting metrics during off-peak hours?
- To identify maximum possible load
- To establish a normal low-load reference point (Correct answer)
- To measure SLA compliance only
- To tune garbage collection settings
Correct answer: To establish a normal low-load reference point
Off-peak baselines capture minimum expected resource use, enabling comparison when load increases.
Question 3: A web service shows high response times but low CPU utilization. Which bottleneck is most likely responsible?
- Insufficient CPU cores
- Memory over-allocation
- I/O or network contention (Correct answer)
- Excessive thread creation
Correct answer: I/O or network contention
High latency with low CPU points to waiting on disk or network rather than compute saturation.
Question 4: What does a 'p99' latency value represent in performance analysis?
- The average latency of all requests
- The latency that 99% of requests fall below (Correct answer)
- The maximum observed latency
- The latency of the 99th request sampled
Correct answer: The latency that 99% of requests fall below
p99 (99th percentile) means 99% of requests complete faster than that value, capturing tail latency.
Question 5: Which tool is best suited for profiling CPU hotspots in a running JVM application without restarting it?
- Wireshark
- async-profiler (Correct answer)
- iostat
- netstat
Correct answer: async-profiler
async-profiler attaches to a live JVM and samples stack traces to identify CPU hotspots.
Question 6: What performance anti-pattern occurs when increasing the number of threads causes throughput to decrease?
- Cache stampede
- Thread starvation
- Coordinated omission
- Contention-induced degradation (Correct answer)
Correct answer: Contention-induced degradation
When threads compete heavily for shared locks or resources, adding more threads increases contention and reduces throughput.
Question 7: In load testing terminology, what does 'think time' simulate?
- Server processing delay
- Network round-trip time
- The pause between user actions (Correct answer)
- Database query execution time
Correct answer: The pause between user actions
Think time models realistic user behavior by inserting pauses between successive requests in a load test.
MeasureUp Test Performance Analysis 2
Which metric directly measures the percentage of time a CPU is actively executing instructions rather than waiting for I/O?