Skip to main content
To test how many requests your self-hosted Kafka setup can handle, you can use benchmarking tools designed for Kafka, such as Kafka’s built-in performance testing tools or third-party tools like kafka-bench, kafka-perf, or Apache JMeter. Here’s a detailed guide on how to perform these tests:

​
Method 1: Using Kafka’s Built-in Performance Tools

Kafka comes with built-in tools that can help you test throughput and performance.

​
Step 1: Start Kafka and Zookeeper

Make sure your Kafka and Zookeeper instances are running. You can check their status using the following commands: 
# Check Zookeeper status
ps -ef | grep zookeeper

# Check Kafka status
ps -ef | grep kafka

​
Step 2: Create a Test Topic

If you haven’t already created a test topic, create one: 
cd /opt/kafka_2.13-3.5.0
bin/kafka-topics.sh --create --topic test-topic --bootstrap-server localhost:9092 --partitions 3 --replication-factor 1

​
Step 3: Use the Kafka Producer Performance Tool

  1. Run the producer performance test to send messages: 
    bin/kafka-producer-perf-test.sh --topic test-topic --num-records 100000 --record-size 100 --throughput 1000 --producer-props bootstrap.servers=localhost:9092
    • --num-records: Total number of records (messages) to send.
    • --record-size: Size of each record in bytes.
    • --throughput: Rate limit for records per second.
  2. Review the output to see how many requests were successfully sent and the throughput achieved.

​
Step 4: Use the Kafka Consumer Performance Tool

  1. Run the consumer performance test to read messages: 
    bin/kafka-consumer-perf-test.sh --topic test-topic --num-messages 100000 --bootstrap-server localhost:9092
    • --num-messages: Total number of messages to consume.
  2. Review the output to see the throughput achieved when consuming messages.

​
Method 2: Using Apache JMeter

Apache JMeter is a popular performance testing tool that can simulate load and measure the throughput of your Kafka setup.

​
Step 1: Install JMeter

  1. Download JMeter from the official website and extract it.
  2. Navigate to the JMeter bin directory: 
    cd /path/to/jmeter/bin

​
Step 2: Configure JMeter for Kafka Testing

  1. Start JMeter: 
    ./jmeter
  2. Create a new Test Plan:
    • Add a Thread Group to simulate multiple users.
    • Add a Kafka Producer Sampler to send messages to your Kafka topic:
      • Set the Kafka Broker to localhost:9092.
      • Specify the topic name.
      • Configure message settings (e.g., message size, number of messages).
  3. Add a Listener to view results:
    • Add a View Results Tree or Aggregate Report to see the test results.

​
Step 3: Run the Test

  • Start the test from JMeter and monitor the throughput and performance metrics in the listeners you’ve set up.

​
Method 3: Using Kafka-bench

Kafka-bench is a simple benchmarking tool that can help you test Kafka performance.
  1. Clone the repository: 
    git clone https://github.com/yourkafka/kafka-bench.git
cd kafka-bench
  2. Compile the code: 
    mvn clean package
  3. Run the benchmark: 
    java -cp target/kafka-bench-1.0-SNAPSHOT.jar com.yourkafka.bench.Benchmark --topic test-topic --num-messages 100000 --num-producers 10 --num-consumers 10
    • Adjust the parameters as needed for your test scenario.

​
Analyzing Results

  • Throughput: Measure the number of messages produced/consumed per second. Higher throughput indicates better performance.
  • Latency: Measure the time taken for messages to be sent and received. Lower latency is preferable.
  • Resource Usage: Monitor CPU, memory, and disk usage on your server while running the tests to identify potential bottlenecks.

​
Conclusion

By using the above methods, you can effectively test how many requests your Kafka setup can handle. Make sure to adjust the parameters according to your testing requirements and the capabilities of your server. After testing, you can analyze the results to determine if your Kafka configuration meets your performance needs or if further tuning is required.