Comprehensive Guide to Performance Testing and Monitoring with JMeter, JConsole, and JVisualVM

This article provides a detailed, step‑by‑step tutorial on performance testing and monitoring of Java backend services.

1. What Is Stress Testing

1.1 Simple Explanation

Stress testing evaluates the maximum load a software and hardware environment can handle and identifies bottlenecks.

1.2 Performance Metrics

Response Time (RT)

Hits Per Second (HPS)

Queries Per Second (QPS)

MaxRT, MinRT, 90% response time

Throughput, error rate, etc.

2. JMeter Performance Testing Tool

2.1 Installation

Download JMeter from https://jmeter.apache.org/download_jmeter.cgi and use version apache-jmeter-5.3 .

2.2 Running JMeter

Execute \apache-jmeter-5.3\bin\jmeter.bat to launch the GUI.

2.3 Creating a Thread Group

Configure 200 threads to start within 1 s, loop 100 times, resulting in 20 000 requests.

2.4 Adding HTTP Request

Set protocol to HTTP, server to www.baidu.com , port 80 .

2.5 Adding Listeners

Include View Results Tree, Summary Report, and Aggregate Report.

2.6 Executing the Test

Start the test, then examine individual request results, summary report (average response time 151 ms, throughput 880 req/s), and aggregate report (median 59 ms, 90% ≤ 271 ms, error rate 0.41%).

3. Performance Monitoring with jconsole

Launch jconsole from the command line, select the passjava-question microservice, and view real‑time metrics such as heap usage, thread count, class loading, and CPU usage.

3.1 Overview

Shows memory, threads, classes, and CPU trends.

3.2 Memory, Threads, Classes, VM Summary, MBean

Detailed charts for each aspect, including loaded/unloaded class counts and VM parameters.

4. Performance Monitoring with jvisualvm

jvisualvm offers richer UI and plugin support.

4.1 Overview

Displays JVM version, startup parameters, and basic metrics.

4.2 Monitoring

Shows CPU, heap, class, and thread usage with more attractive charts.

4.3 Thread Details

Lists thread states: running, sleeping, waiting, parked, and blocked.

4.4 Sampler

Allows sampling of CPU or memory usage.

4.5 Plugins (Visual GC)

Install via Tools → Plugins → Visual GC to visualize garbage‑collection activity.

5. Gateway Performance Testing

Configure JMeter to target the API Gateway at port 8060 , send 200 requests per second continuously, and observe throughput of 2422 req/s, 90% response time 142 ms, 99% response time 281 ms.

6. Microservice Performance Testing

Add a simple test endpoint to passjava-question , verify with Postman or browser, then run JMeter against it, achieving throughput 3542 req/s, 90% response time 100 ms, 99% response time 152 ms.

7. Combined Gateway + Microservice Testing

Test the full request chain via http://localhost:8060/api/question/v1/admin/question/test , resulting in throughput 982 req/s, 90% response time 437 ms, 99% response time 790 ms, demonstrating a performance drop due to the additional network hop.

8. Database Query Optimization

Measure query latency for http://localhost:11000/question/v1/admin/question/list?type=5 . Initial latency 43 ms; after adding a regular index on the type column, latency drops to 18 ms.

9. Garbage‑Collection Optimization

Observe frequent GC in the Eden space (32 MB) causing 480 stop‑the‑world events per minute and low throughput (275 req/s). Increase heap to -Xmx1024m -Xms1024m -Xmn512m , reducing GC events to 92 per minute and raising throughput to 347 req/s.

10. Summary

The article demonstrates how to use JMeter for load testing, jconsole and jvisualvm for monitoring, and practical optimization techniques such as reducing middleware hops, adding database indexes, and enlarging the Eden space to improve Java application performance.