Performance Testing Metrics

• A deep understanding of performance testing metrics not only helps comprehensively evaluate system performance but also provides reliable data for optimization decisions.
• This guide systematically introduces the core performance metrics of AngusTester, helping you quickly grasp the meaning and usage of each metric.

Basic Counting Metrics

Quantify the core operations and data flow during testing, providing foundational data for performance analysis.

Metric Name	Parameter Name	Description	Key Purpose	Calculation Logic
Iteration Count	iterations	Number of test task executions	Measure test coverage	Accumulated execution count
Sample Count	n	Number of valid data collection points	Evaluate data validity	Count valid sample points
Operation Count	operations	Total number of request operations executed	Assess system processing capacity	Accumulated operation count
Transaction Count	transactions	Number of completed business transactions	Evaluate business processing capacity	Accumulated successful transactions
Bytes Read	readBytes	Total data read	Assess network/storage read load	Accumulated bytes read
Bytes Written	writeBytes	Total data written	Assess network/storage write load	Accumulated bytes written

💡 Application Tip: Basic metrics must be analyzed with a time dimension to reveal performance characteristics.

Throughput Metrics

Key indicators reflecting the system's processing capacity per unit of time.

Request Throughput

Metric Name	Parameter Name	Description	Key Purpose	Calculation Formula
Operations Per Second	ops	Number of requests processed per second	Evaluate system processing capacity	Operation count / Test duration
Transactions Per Second	tps	Number of transactions completed per second	Evaluate business processing capacity	Transaction count / Test duration

💡 Important Note: In single-interface testing scenarios, QPS test results and TPS values are equal.

Network Throughput

Metric Name	Parameter Name	Description	Key Purpose	Calculation Formula
Bytes Read Per Second	brps	Data read per second	Evaluate network receiving capacity	Bytes read / Test duration
Bytes Written Per Second	bwps	Data written per second	Evaluate network sending capacity	Bytes written / Test duration

⚠️ Critical Threshold: When throughput reaches 80% of the system's theoretical peak, it may indicate an approaching performance bottleneck.

Latency Metrics

Core indicators measuring system response speed, directly impacting user experience.

Metric Name	Parameter Name	Description	Key Purpose	Calculation Logic
Average Response Time	tranMean	Average transaction response time	Reflect overall system performance	Total response time / Transaction count
Minimum Response Time	tranMin	Best response time	Show optimal performance	Take minimum value
Maximum Response Time	tranMax	Worst response time	Reveal performance bottlenecks	Take maximum value
P50 Response Time	tranP50	Response time for 50% of requests	Reflect typical response performance	50th percentile after sorting
P75 Response Time	tranP75	Response time for 75% of requests	Reflect good performance	75th percentile after sorting
P90 Response Time	tranP90	Response time for 90% of requests	Evaluate high-quality performance	90th percentile after sorting
P95 Response Time	tranP95	Response time for 95% of requests	Evaluate high service level	95th percentile after sorting
P99 Response Time	tranP99	Response time for 99% of requests	Evaluate extreme-case performance	99th percentile after sorting
P999 Response Time	tranP999	Response time for 99.9% of requests	Evaluate ultimate performance	99.9th percentile after sorting

Error Metrics

Core indicators for evaluating system stability and reliability.

Metric Name	Parameter Name	Description	Key Purpose	Calculation Formula
Error Count	errors	Number of errors occurred	Evaluate system stability	Accumulated error count
Error Rate	errorRate	Proportion of errors occurred	Evaluate system reliability	(Error count / Operation count) × 100%
Error Cause Distribution	errorCauseCounter	Distribution of error types	Diagnose root causes	Group statistics by error type

🔍 Diagnostic Tip: Connection timeouts and 5xx errors should be prioritized in error types.

Thread (Concurrency) Metrics

Reflect system concurrency processing capability and resource utilization efficiency.

Metric Name	Parameter Name	Description	Key Purpose	Status Description
Thread Pool Size	threadPoolSize	Current thread pool capacity	Evaluate system concurrency capacity	-
Active Thread Count	threadPoolActiveSize	Number of working threads	Evaluate resource utilization	-
Maximum Thread Pool Capacity	threadMaxPoolSize	Maximum supported threads	Evaluate system expansion limit	-
Thread Running Status	threadRunning	Whether the thread is running	Monitor thread status	true = running
Thread Termination Status	threadTerminated	Whether the thread is terminated	Monitor thread status	true = terminated

Timestamp Fields

Field Name	Parameter Name	Description
Server Timestamp	timestamp	Server-recorded time
Runner Timestamp	timestamp0	Sampling task-recorded time
Sampling Task Name	name	Sampling task identifier
Test Duration	duration	Total execution time
Sampling Duration	duration0	Single sampling time
Start Time	startTime	Test start time point
End Time	endTime	Test end time point

Key Considerations

1. Metric Correlation: When latency increases, synchronously check error rate and throughput changes.
2. Environmental Factors:
   • Network latency directly affects response time.
   • Test data scale impacts throughput performance.
3. Scenario Differences:
   • E-commerce systems should focus on P99 latency.
   • Financial systems must ensure 0% error rate.
4. Monitoring Strategy:
   • Establish performance baselines for comparison.
   • Set threshold-based alerts (e.g., error rate > 0.5%).
5. Trend Analysis: Compare with historical data to identify performance degradation.