I85PREP
⌘ K
ISTQB CT-PT study materials
ISTQB CT-PT study resources

ISTQB CT-PT practice questions

One question at a time. Build your ISTQB CT-PT understanding.

First 50 of 200 questions. Each one lists the choices, the correct answer, and an explanation. The checker at the top of this page saves progress in this browser when it is available.

Continue in the practice tool above to load more questions, 50 at a time.

Question 1Performance Testing Fundamentals (basic concepts)

In performance testing, what does response time measure?

  1. The elapsed time from sending a request until the complete response is received Correct answer.
  2. The volume of data written to disk during a test run
  3. The number of virtual users active at any one moment
  4. The percentage of requests that return an error

Explanation

Response time is the end-to-end latency experienced by a request, from dispatch to full receipt of the reply. It is the primary user-facing metric and is usually reported as an average plus percentiles such as the 90th or 95th. Throughput, error rate, and resource consumption are separate metric families.

Question 2Performance Testing Fundamentals (basic concepts)

Which metric expresses the amount of work a system completes per unit of time during a performance test?

  1. Throughput Correct answer.
  2. Utilization
  3. Think time
  4. Latency

Explanation

Throughput counts completed transactions, requests, or bytes per unit time, for example requests per second. It shows capacity independent of how long any single request took. Latency and utilization are timing and resource measures respectively and do not express completed work per second.

Question 3Performance Testing Fundamentals (basic concepts)

What is the purpose of a performance test baseline?

  1. To replace the need for production monitoring
  2. To prove the system can never fail under peak load
  3. To measure only the developer's local machine speed
  4. To record a reference set of results against which later builds are compared Correct answer.

Explanation

A baseline is a captured, repeatable set of measurements from a known build and environment. Later runs are compared against it to detect regressions or confirm improvements. It does not guarantee freedom from failure and does not substitute for live monitoring.

Question 4Performance Testing Fundamentals (basic concepts)

A virtual user in a load testing tool most closely represents which of the following?

  1. A simulated concurrent end user issuing a scripted business transaction Correct answer.
  2. A database index used to speed up queries
  3. A network switch forwarding packets
  4. A physical server in the test environment

Explanation

A virtual user emulates the behavior of a real user by replaying a scripted sequence of requests, often with think time and pacing. Many virtual users run concurrently to generate load. Servers, indexes, and switches are infrastructure components, not simulated users.

Question 5Performance Testing Fundamentals (basic concepts)

Which term describes the maximum load a system can handle while still meeting its defined performance objectives?

  1. Reliability
  2. Capacity Correct answer.
  3. Maintainability
  4. Scalability

Explanation

Capacity is the ceiling of load the system sustains while still satisfying agreed service-level objectives such as response time and error rate. Scalability instead describes how capacity changes when resources are added. Reliability concerns failure behavior over time.

Question 6Performance Testing Fundamentals (basic concepts)

In performance testing terminology, what is think time?

  1. The time a server spends thinking before returning an error
  2. The time required to compile the test script
  3. The interval between two scheduled test executions
  4. The simulated pause a virtual user waits between consecutive requests Correct answer.

Explanation

Think time models the real user's pause to read a page or decide the next action, and it reduces the request rate per virtual user. It is configured in the load script and directly affects generated throughput. It does not refer to server processing, script compilation, or scheduling gaps.

Question 7Performance Testing Fundamentals (basic concepts)

Which performance test type is designed to verify system behavior at or just beyond the expected peak load?

  1. Load test
  2. Unit test
  3. Stress test Correct answer.
  4. Smoke test

Explanation

A stress test pushes load to or beyond expected peaks to observe degradation, error handling, and recovery. A load test instead validates behavior at anticipated normal-to-peak levels. Smoke and unit tests check basic functionality, not behavior under extreme load.

Question 8Performance Testing Fundamentals (basic concepts)

Which of the following best describes a soak test?

  1. A sustained moderate load held for an extended period to reveal degradation over time Correct answer.
  2. A test that measures only the first request after startup
  3. A short burst at maximum load to find the breaking point
  4. A test executed without any monitoring enabled

Explanation

A soak or endurance test applies a steady, realistic load for hours or days to expose memory leaks, resource exhaustion, and gradual performance drift. It differs from a stress test, which seeks the breaking point quickly. Short single-request runs cannot reveal long-term degradation.

Question 9Performance Testing Fundamentals (basic concepts)

A test reports a 95th percentile response time of 2 seconds. How should this be interpreted?

  1. The average response time was 2 seconds
  2. Only 5 percent of requests succeeded
  3. Exactly 95 requests took 2 seconds each
  4. Ninety-five percent of requests completed in 2 seconds or less Correct answer.

Explanation

The 95th percentile is the value below which 95 percent of measured response times fall. It highlights the experience of slower requests that an average would hide. It says nothing about absolute counts, the mean, or success rates.

Question 10Performance Testing Fundamentals (basic concepts)

Why is the arithmetic mean response time alone considered misleading when reporting performance results?

  1. It hides the slow tail of requests that many users actually experience Correct answer.
  2. It cannot be calculated for web transactions
  3. It only applies to batch jobs
  4. It always equals the median value

Explanation

Averages smooth out outliers, so a small number of very slow requests can be invisible while still harming real users. Reporting percentiles alongside the mean exposes the tail behavior. The mean is computable for web transactions and generally differs from the median.

Question 11Performance Testing Fundamentals (basic concepts)

During a load test, throughput plateaus while response time rises sharply as virtual users increase. What does this pattern most likely indicate?

  1. The test tool has stopped generating requests
  2. Think time was set too high
  3. Response time is improving with load
  4. The system has reached a bottleneck and is saturating Correct answer.

Explanation

When added load no longer increases completed work but latency climbs, a resource has become the limiting factor and the system is saturating. This is the classic approach to the capacity ceiling. A stopped generator would reduce load, and higher think time would lower request rates rather than raise latency.

Question 12Performance Testing Fundamentals (basic concepts)

What is the main risk of running a performance test without a representative production-like environment?

  1. Results may not predict production behavior because capacity and configuration differ Correct answer.
  2. Percentiles cannot be computed
  3. The test script cannot be recorded
  4. Virtual users cannot be configured

Explanation

Hardware, network topology, data volume, and configuration strongly influence performance, so results from a scaled-down or misconfigured environment may mislead. Testers must document environment differences and interpret results cautiously. Script recording, user configuration, and percentile calculation remain possible regardless of environment fidelity.

Question 13Performance Testing Fundamentals (basic concepts)

Which activity belongs to performance test design rather than performance test execution?

  1. Monitoring CPU during the run
  2. Stopping the run after a threshold breach
  3. Defining the workload mix and load profile to be applied Correct answer.
  4. Collecting response time samples

Explanation

Design activities decide what will be tested and how load will be shaped, including the workload mix, load profile, and acceptance criteria. Monitoring, sample collection, and run control happen during execution. Separating design from execution keeps tests purposeful and repeatable.

Question 14Performance Testing Fundamentals (basic concepts)

A team must validate that a system recovers gracefully after an unexpected loss of a database node during peak traffic. Which test type fits best?

  1. Failover or resilience test Correct answer.
  2. Soak test
  3. Configuration test
  4. Spike test

Explanation

Resilience testing injects faults such as node loss under load and verifies that the system fails over and recovers within acceptable limits. Soak tests target endurance, and spike tests target sudden load surges. Configuration testing checks settings rather than fault recovery.

Question 15Performance Testing Fundamentals (basic concepts)

What distinguishes a spike test from a stress test?

  1. A spike test applies a sudden, short surge in load, while a stress test gradually pushes beyond capacity Correct answer.
  2. A spike test is always run in production
  3. A stress test only measures disk usage
  4. A spike test uses no virtual users

Explanation

Spike testing injects an abrupt, brief increase in traffic to check how the system absorbs and recovers from sudden surges. Stress testing instead ramps load progressively to find the breaking point and observe degradation. Both use virtual users and neither is defined by disk metrics or production-only execution.

Question 16Performance Testing Fundamentals (basic concepts)

Which combination of metrics best supports a conclusion that a bottleneck is CPU-bound?

  1. Low CPU utilization with rising response time
  2. Sustained high CPU utilization that coincides with throughput plateau and rising latency Correct answer.
  3. High think time with low error rate
  4. Increasing disk free space during the run

Explanation

A CPU bottleneck is evidenced when CPU saturates at the same point throughput stops growing and latency climbs. Correlation across metrics is essential to attribute the limit to the processor. Low CPU with rising latency points elsewhere, and think time or disk free space do not identify a CPU limit.

Question 17Performance Testing Fundamentals (basic concepts)

What is the primary role of workload modeling in performance testing?

  1. To define a realistic mix and volume of transactions that mirrors expected usage Correct answer.
  2. To decide the color scheme of reports
  3. To replace the need for acceptance criteria
  4. To select the programming language of the test script

Explanation

Workload modeling determines which transactions occur, in what proportions, and at what rates so the generated load reflects real usage. An unrealistic mix produces results that cannot guide decisions. It does not concern scripting language, reporting cosmetics, or acceptance criteria.

Question 18Performance Testing Fundamentals (basic concepts)

A test harness on a single machine shows high CPU on the load generator itself while the target server is lightly loaded. What is the most appropriate interpretation?

  1. The generator is a bottleneck and results are invalid for server capacity Correct answer.
  2. The server has reached its capacity limit
  3. Response time percentiles are unnecessary
  4. The network is definitely the cause

Explanation

If the load generator saturates, it cannot produce the intended request rate, so measured server behavior does not reflect the target load. The fix is to distribute generation or reduce per-generator overhead before drawing capacity conclusions. A lightly loaded server clearly has not hit its own limit.

Question 19Performance Testing Fundamentals (basic concepts)

Which statement best describes the relationship between load and response time in the early, linear region of a load curve?

  1. Response time stays flat regardless of load
  2. Response time grows slowly and predictably as load increases Correct answer.
  3. Response time is undefined until saturation
  4. Response time falls as load increases

Explanation

Before contention appears, added load causes only modest, roughly proportional increases in response time, forming the linear region of the curve. The knee occurs later when a resource saturates and latency rises steeply. Response time does not decrease with load or become undefined.

Question 20Performance Testing Fundamentals (basic concepts)

Why should performance acceptance criteria be defined before test execution begins?

  1. Because tools cannot run without them
  2. To allow the thresholds to be changed after seeing results
  3. So results can be judged objectively against agreed thresholds Correct answer.
  4. To eliminate the need for monitoring

Explanation

Predefined criteria such as target response times and error budgets let stakeholders judge pass or fail objectively and prevent post-hoc rationalization. They also guide test design and environment sizing. They do not remove the need for monitoring and must not be retrofitted to match results.

Question 21Performance Testing Fundamentals (basic concepts)

A system passes a short load test at 500 concurrent users but fails after two hours at the same load. Which explanation is most consistent with this observation?

  1. The workload mix changed between runs
  2. Think time was removed
  3. The percentile calculation was wrong
  4. A slow resource leak accumulates until exhaustion Correct answer.

Explanation

Stable load that degrades over hours typically points to a gradual leak of memory, connections, or handles that eventually exhausts a resource. This is exactly what soak testing is designed to reveal. A changed workload mix, calculation error, or removed think time would not produce time-dependent failure at constant load.

Question 22Performance Testing Fundamentals (basic concepts)

When interpreting a load curve, what does the 'knee' point represent?

  1. The first virtual user added to the test
  2. The load level where the system transitions from stable to degraded performance Correct answer.
  3. The point where think time equals zero
  4. The maximum number of script iterations

Explanation

The knee marks where response time or resource use begins to rise steeply for small increases in load, signaling imminent saturation. Operating comfortably below the knee provides headroom for traffic variation. It is unrelated to user counts, iteration limits, or think time configuration.

Question 23Performance Testing Fundamentals (basic concepts)

Two test runs produce identical average response times, but run A has a much wider spread with a heavy slow tail. What is the most defensible conclusion?

  1. The two runs are equivalent because averages match
  2. Run B must have more errors
  3. Run A likely delivers a worse user experience despite the same mean Correct answer.
  4. Run A used fewer virtual users

Explanation

Equal means can mask very different distributions; a heavy tail means more users endure slow responses even though the average is unchanged. Percentile and distribution analysis is needed to distinguish the runs. Error counts and user counts are not implied by the timing spread.

Question 24Performance Testing Fundamentals (basic concepts)

A team plans to test a system whose production traffic follows a daily peak at noon and a nightly batch window. Which approach best reflects this in the load profile?

  1. Only a spike test at midnight
  2. A soak test at minimal load
  3. A single constant load for the entire test
  4. A load profile that models the daily peak and the batch overlap explicitly Correct answer.

Explanation

Modeling the actual temporal pattern, including the noon peak and the batch overlap, exposes contention that a flat profile would hide. The load profile should reproduce realistic timing and mix so results predict production behavior. Constant load, a single midnight spike, or minimal-load endurance runs miss the key interactions.

Question 25Performance Testing Fundamentals (basic concepts)

Which factor most directly explains why results from a test using synthetic data differ from those using production-like data volumes?

  1. Production data changes the scripting language
  2. Test tools forbid synthetic data
  3. Synthetic data always uses more virtual users
  4. Query plans and cache behavior depend on data size and distribution Correct answer.

Explanation

Database optimizers choose plans based on statistics, and caches hit or miss depending on data volume and distribution, so small synthetic datasets can produce unrealistically fast results. Using representative volumes and distributions improves validity. Virtual user counts, scripting language, and tool rules are unaffected by data realism.

Question 26Performance Testing Fundamentals (basic concepts)

In the ISTQB performance testing vocabulary, what does a 'virtual user' represent during a load test?

  1. A member of the development team who manually operates the system during testing
  2. A physical server instance provisioned to absorb the generated traffic
  3. A simulated concurrent end user whose actions are scripted and replayed by the load generator Correct answer.
  4. A monitoring agent installed on the application server to collect metrics

Explanation

A virtual user is an emulated end user created by the load generation tool, executing a scripted business scenario so that realistic concurrent load can be applied without real people. Virtual users are logical constructs of the tool, not humans and not servers, and they are distinct from monitoring agents that observe the system under test.

Question 27Performance Testing Fundamentals (basic concepts)

Which statement correctly distinguishes load testing from stress testing?

  1. Load testing validates behavior at expected load levels, while stress testing pushes beyond normal capacity to find limits and failure behavior Correct answer.
  2. Load testing is always automated, while stress testing is always performed manually by testers
  3. Load testing targets the database tier, while stress testing targets the network tier
  4. Load testing measures only response time, while stress testing measures only resource utilization

Explanation

Load testing verifies that the system meets its requirements under anticipated or typical load, whereas stress testing deliberately exceeds normal capacity to discover breaking points, degradation patterns, and recovery behavior. The two differ in the load level and the intent of the investigation, not in tooling, metrics, or architecture tier.

Question 28Performance Testing Fundamentals (basic concepts)

What is the purpose of a baseline performance test?

  1. To record a reference set of measurements against which later builds or releases can be compared Correct answer.
  2. To determine the absolute maximum number of users the system can ever support
  3. To prove that the production environment is unnecessary for testing
  4. To verify that functional requirements are implemented correctly

Explanation

A baseline establishes a known, repeatable set of measurements, usually on a stable build and environment, so that future results can be compared to detect regressions or improvements. It is a reference point, not an attempt to find the absolute capacity ceiling, which is the role of stress or capacity testing.

Question 29Performance Testing Fundamentals (basic concepts)

Which metric describes the number of completed business transactions or requests a system processes per unit of time?

  1. Utilization
  2. Throughput Correct answer.
  3. Think time
  4. Latency

Explanation

Throughput expresses how much work the system completes per unit time, such as transactions per second or requests per second. Latency measures delay for a single request, think time is a modeled pause between user actions, and utilization describes how busy a resource is.

Question 30Performance Testing Fundamentals (basic concepts)

What does response time measure in a performance test?

  1. The elapsed time between sending a request and receiving the complete response Correct answer.
  2. The time a server's CPU spends executing the request handler
  3. The interval a virtual user waits before issuing the next request
  4. The total duration of the entire test run

Explanation

Response time is the elapsed time from issuing a request to receiving the full response, including network and server processing contributions. It is a per-transaction measure, distinct from CPU execution time, from the modeled think time, and from the overall duration of the test campaign.

Question 31Performance Testing Fundamentals (basic concepts)

Which of the following is a typical goal of performance testing?

  1. Assessing whether the system meets agreed responsiveness and capacity objectives under representative load Correct answer.
  2. Ensuring that source code follows naming conventions
  3. Confirming that every code branch is executed at least once
  4. Verifying that user interface labels match the specification

Explanation

Performance testing evaluates responsiveness, throughput, resource usage, and scalability against defined objectives under representative load conditions. Code coverage, UI label checks, and naming conventions are functional or static quality concerns unrelated to performance characteristics.

Question 32Performance Testing Fundamentals (basic concepts)

What is a workload profile in performance testing?

  1. A schedule of which testers run which scripts
  2. A list of the hardware specifications of the servers under test
  3. A description of the mix, volume, and timing of user activities that the test will reproduce Correct answer.
  4. A chart of CPU utilization recorded during a previous test

Explanation

A workload profile characterizes the expected user population and behavior, including the distribution of business transactions, concurrency levels, and arrival patterns, so the test reproduces realistic demand. It is not a hardware inventory, a monitoring chart, or a staffing plan.

Question 33Performance Testing Fundamentals (basic concepts)

Which term describes the degree to which a system can handle growing amounts of work by adding resources?

  1. Reliability
  2. Usability
  3. Maintainability
  4. Scalability Correct answer.

Explanation

Scalability is the ability of a system to sustain growth in workload through added resources such as servers, CPU, or memory, ideally with proportional performance gains. Reliability concerns consistent correct operation over time, maintainability concerns ease of change, and usability concerns ease of human interaction.

Question 34Performance Testing Fundamentals (basic concepts)

A test team must estimate how many concurrent users the system can support while keeping the 95th percentile response time under 3 seconds. Which performance test type is most appropriate?

  1. A failover test that shuts down a node during operation
  2. Capacity testing, increasing load in steps until the objective threshold is reached Correct answer.
  3. A configuration test comparing two application server settings
  4. A soak test held at a fixed low load for many hours

Explanation

Capacity testing incrementally raises load to determine the maximum workload the system can sustain while still meeting a defined performance objective such as a percentile response time threshold. Soak, failover, and configuration tests answer different questions about endurance, resilience, and settings, not the supported concurrency ceiling.

Question 35Performance Testing Fundamentals (basic concepts)

A system must be shown to degrade gracefully under sudden, brief surges far above normal traffic, such as a flash sale announcement. Which test type addresses this requirement most directly?

  1. A baseline test that records a single measurement at normal load
  2. A soak test that holds steady load for an extended period
  3. A functional regression test of the checkout workflow
  4. A spike test that applies an abrupt, large increase in load and then returns to normal Correct answer.

Explanation

A spike test reproduces a sudden, sharp jump in load followed by a return to normal, verifying that the system survives and recovers rather than collapsing. Baseline and soak tests use steady load levels, and functional regression tests do not exercise surge behavior.

Question 36Performance Testing Fundamentals (basic concepts)

Why should a performance test script include parameterized data and realistic think times rather than fixed values and zero pauses?

  1. To ensure the test can run without a network connection
  2. To reproduce realistic user pacing and data variety, avoiding artificial cache hits and unrealistically high load Correct answer.
  3. To guarantee that the test completes faster than a manual test would
  4. To eliminate the need for any monitoring during execution

Explanation

Parameterized data avoids every user requesting identical records, which would produce cache hits and unrealistically low database load, while think times model the pauses between real user actions and control the resulting request rate. Together they make the generated workload representative of production behavior.

Question 37Performance Testing Fundamentals (basic concepts)

A performance test measures a 90th percentile response time of 1.5 seconds and a maximum of 12 seconds. What is the most defensible interpretation?

  1. The system failed to meet any response time objective
  2. Every request completed within 1.5 seconds
  3. The average response time was exactly 12 seconds
  4. Most requests were fast, but a small fraction experienced severe delays that merit investigation Correct answer.

Explanation

The 90th percentile indicates that 90 percent of requests completed at or below 1.5 seconds, while the 12-second maximum reveals a slow tail affecting a minority of transactions. This combination suggests generally good responsiveness with outliers that may indicate contention, garbage collection, or timeouts worth diagnosing.

Question 38Performance Testing Fundamentals (basic concepts)

During a load test, the load generator's own CPU reaches saturation while the target server remains lightly utilized. What is the most appropriate corrective action?

  1. Increase the number of virtual users on the same generator to push harder
  2. Reduce monitoring to free resources on the target server
  3. Distribute load generation across multiple machines and re-run before drawing conclusions Correct answer.
  4. Conclude that the application has a severe performance defect

Explanation

When the load generator saturates, it can no longer produce the intended load reliably, so measured response times reflect the generator's limits rather than the system under test. Spreading generation across multiple machines removes that constraint and yields trustworthy results.

Question 39Performance Testing Fundamentals (basic concepts)

Which statement best describes the closed workload model used by most load testing tools?

  1. Requests arrive at a fixed rate independent of how slowly the system responds
  2. No virtual users are simulated and traffic is replayed from captured packets only
  3. A fixed set of virtual users each waits for a response before issuing its next request Correct answer.
  4. Load is generated only by real users while testers observe the system

Explanation

In a closed model a fixed population of virtual users cycles through think time, request, and wait for response, so slower responses reduce the request rate. This self-throttling behavior differs from open models, where requests arrive at a defined rate regardless of system responsiveness.

Question 40Performance Testing Fundamentals (basic concepts)

A team must decide how many virtual users to configure for a test of a system expected to serve 10,000 sessions per hour with an average of 60 seconds of think time per user action. Which input is essential to derive the concurrency level?

  1. The programming language used by the application
  2. The number of servers in the production cluster
  3. The transaction arrival rate and the average time each virtual user spends per iteration Correct answer.
  4. The size of the test data files on disk

Explanation

Concurrency is derived from arrival rate and the time a virtual user occupies the system per iteration, following the relationship between throughput, concurrency, and think time. Server count, language, and data file size do not determine how many simulated users are needed to reproduce the target request rate.

Question 41Performance Testing Fundamentals (basic concepts)

What is the main purpose of performance monitoring during a load test?

  1. To correlate system resource and application metrics with observed response times and identify bottlenecks Correct answer.
  2. To replace the need for a defined workload model
  3. To measure the functional correctness of business rules
  4. To automatically fix configuration problems found during the run

Explanation

Monitoring collects CPU, memory, disk, network, and application-level metrics that, when correlated with response times and throughput, reveal where the system is constrained. It does not substitute for workload modeling, repair problems automatically, or validate business rules.

Question 42Performance Testing Fundamentals (basic concepts)

A test uses a production-like environment but only 5 percent of the production data volume. Which risk most directly affects the validity of the results?

  1. Database query performance and cache behavior may differ substantially from production Correct answer.
  2. The test will automatically become a stress test
  3. The response time objective can no longer be defined
  4. The load generator will be unable to create any virtual users

Explanation

Data volume strongly influences query plans, index selectivity, cache hit ratios, and I/O patterns, so a small dataset can make the system appear faster than it will be in production. Virtual user creation, objective definition, and test classification are unaffected by data volume.

Question 43Performance Testing Fundamentals (basic concepts)

Which of the following best explains why a performance test should be repeated to confirm a suspected bottleneck?

  1. A single run is always sufficient for any performance conclusion
  2. Run-to-run variance from caches, background jobs, and shared infrastructure can produce misleading single results Correct answer.
  3. Repeated runs eliminate the need to monitor resource utilization
  4. Repeating a test changes the application code under test

Explanation

Environmental noise such as warm caches, concurrent batch jobs, and shared infrastructure can shift results between runs, so repeating a test helps separate genuine bottlenecks from incidental variation. Repetition does not alter the code, and monitoring remains necessary to locate the cause.

Question 44Performance Testing Fundamentals (basic concepts)

A performance test reveals that response time stays flat as load increases, then rises steeply beyond a certain point. What does the steep rise most likely indicate?

  1. The load generator has stopped sending requests entirely
  2. The response time objective has been automatically relaxed
  3. The business transactions have become functionally incorrect
  4. The system has reached a resource saturation point where queueing delays grow rapidly Correct answer.

Explanation

The transition from a flat region to a steep rise marks saturation of a constrained resource, after which additional load is absorbed as queueing and latency grows disproportionately. It is not evidence that the generator stopped, that objectives changed, or that functional correctness failed.

Question 45Performance Testing Fundamentals (basic concepts)

Which activity is most characteristic of performance test planning rather than performance test analysis?

  1. Comparing measured percentiles against the agreed thresholds
  2. Correlating CPU saturation with the response time curve
  3. Identifying the resource that limits throughput
  4. Defining objectives, workload model, environment, and success criteria for the test Correct answer.

Explanation

Planning establishes what will be tested, under which workload and environment, and against which criteria before execution begins. Correlating metrics, locating limiting resources, and comparing results to thresholds are analysis activities performed on the collected data.

Question 46Performance Testing Fundamentals (basic concepts)

A retail system experiences a sharp drop in throughput and a rise in errors once concurrent users exceed 800, but recovers fully when load is reduced. Which conclusion is best supported?

  1. The load generator was misconfigured from the start
  2. The system has a capacity limit near 800 users and lacks graceful degradation beyond it Correct answer.
  3. The test data was insufficient to exercise the database
  4. The system has a memory leak that only appears over long durations

Explanation

A repeatable drop in throughput with rising errors at a specific concurrency, followed by recovery when load falls, points to a capacity ceiling and poor behavior beyond it rather than to endurance or data issues. A memory leak would typically manifest as gradual degradation over time at a steady load.

Question 47Performance Testing Fundamentals (basic concepts)

Two runs use the same script and environment. Run A has 500 virtual users with 5 seconds of think time; Run B has 500 virtual users with 30 seconds of think time. What is the most likely effect on the generated request rate?

  1. Run A will fail because short think times are invalid
  2. Both runs will generate identical request rates because user counts match
  3. Run B will generate a lower request rate because each user pauses longer between requests Correct answer.
  4. Run B will generate a higher request rate because users wait longer

Explanation

In a closed workload model the request rate depends on both the number of users and the time each user spends per iteration, so longer think time reduces the rate for the same user count. Equal user counts do not imply equal load when pacing differs, and short think time is a legitimate configuration.

Question 48Performance Testing Fundamentals (basic concepts)

A system meets its response time objective at 1,000 users but the database connection pool is fully utilized at that level. What is the most significant risk for production?

  1. The database will automatically provision additional connections without limit
  2. The connection pool has no influence on performance
  3. Response times will improve as more users arrive
  4. Any further increase in concurrency will cause requests to queue for connections and response times to degrade sharply Correct answer.

Explanation

A fully utilized connection pool is a saturated resource, so additional concurrency forces requests to wait for a free connection, producing queueing and rapid latency growth. Pools have configured maximums rather than unlimited auto-growth, and contention worsens rather than improves performance as load rises.

Question 49Performance Testing Fundamentals (basic concepts)

A team observes that results from a test environment with a fast local network differ greatly from production, where clients connect over high-latency links. Which factor best explains the discrepancy?

  1. Think time is ignored when latency is high
  2. Virtual users cannot be generated over high-latency links
  3. Network latency adds to every round trip and can dominate response time for chatty protocols Correct answer.
  4. The application server executes business logic more slowly on fast networks

Explanation

When a protocol requires many round trips, per-request network latency accumulates and can become a major share of end-to-end response time, which a low-latency test environment hides. Server logic speed, virtual user generation, and think time modeling are not inherently altered by link latency.

Question 50Performance Testing Fundamentals (basic concepts)

A performance test shows high response times while CPU, memory, disk, and network on the application server are all low. Which explanation is most plausible?

  1. The disk subsystem is saturated by the application
  2. The application server is CPU-bound and needs more cores
  3. The network interface is fully utilized
  4. Requests are waiting on an external dependency such as a downstream service or a lock Correct answer.

Explanation

Low utilization across all local resources while latency is high suggests the time is spent waiting rather than computing, for example on a slow downstream call, a database lock, or a connection wait. CPU, disk, and network saturation would each show elevated utilization on the corresponding resource.