Performance Testing with wrk: Analyzing Latency and Throughput

Performance Testing with wrk: Analyzing Latency and Throughput

Introduction:

In the realm of web development, ensuring optimal performance of your application is paramount. Performance testing tools like wrk allow developers to simulate real-world traffic and evaluate how their application performs under various load conditions. In this experiment, we'll delve into using wrk to analyze the latency and throughput of a web application.

Experiment:

Before diving into the experiment, let's set up wrk. If you haven't already installed it, follow these steps:

bash
sudo apt-get install build-essential libssl-dev git -y 
git clone https://github.com/wg/wrk.git wrk

cd wrk

make

sudo cp wrk /usr/local/bin

Now, let's execute a test using wrk:

bashe
wrk -t10 -c1000 -d10s -T30s --latency 'http://10.1.0.20:5000/parse?q=hello&project=5e1717fb0bdf66011ef82702'

Explanation:

-t10: Specifies 10 threads to be used for the test.

-c1000: Configures 1000 connections to be made concurrently.

-d10s: Sets the duration of the test to 10 seconds.

-T30s: Sets a 30-second timeout for each connection.

--latency: Enables latency measurement.

Results:

Thread Stats:

Average latency: 2.01s

Standard deviation: 980.91ms

Maximum latency: 6.36s

Requests/Sec:

Average: 48.45

Maximum: 290.00

Latency Distribution:

50% of requests were served in 2.01s

75% in 2.21s

90% in 3.23s

99% in 5.62s

Overall:

3247 requests were made in 10.09s

Total data transferred: 1.69MB

Average throughput: 321.77 requests/sec

Average transfer rate: 171.57KB/sec

The wrk tool provides valuable insights into the performance of our web application. In this experiment, we observed the latency distribution, average throughput, and request rates under a specific load scenario. By analyzing these metrics, developers can identify potential bottlenecks and optimize their application for better performance.