HTTP Load Tester

HTTP Load Tester is a Python library for load testing and benchmarking HTTP endpoints.

Features

• Load Testing: Send a specified number of HTTP requests to a given URL and measure response times and error rates.
• Stress Testing: Determine the maximum capacity of requests before system failure by gradually increasing the load.
• Rate Limiting: Control the rate of queries per second (QPS) to simulate real-world scenarios. You can use a --qps flag to generate requests at a given fixed QPS.

The inspiration for my implementation of stress testing (determining the maximum capacity of requests before system breaks down) is from https://medium.com/@rico098098/load-testing-with-python-fea13369af43.

The code in load_tester.py utilizes the requests, argparse, and time libraries. It defines a decorator named measure_time to measure and print the execution time of a function. The RateLimitedSession class extends requests.Session to introduce a delay between requests to maintain a desired rate of queries per second (QPS). It overrides the send() method to implement this functionality. The reason I create a requests.Session object to manage and persist settings across multiple requests is because if there are "several requests to the same host, the underlying TCP connection will be reused, which can result in a significant performance increase" (from https://requests.readthedocs.io/en/latest/user/advanced/#session-objects). The LoadTester class initializes with a URL and QPS, and it includes methods to send HTTP requests and run load or stress tests. The main() function parses command-line arguments using argparse and creates a LoadTester object based on the provided arguments, running either a load or stress test based on the specified test type.

There is also code in test_load_tester.py to create a LoadTester instance to run a load test in test_run_load_test() (for instance having the URL be Google and defining a mock user behavior function to search "Fireworks AI" in Google). This test file also includes code to run a stress test in test_run_stress_test() that creates a LoadTester instance to do stress testing to determine the maximum capacity of requests before system failure.

Setup and Usage

To set up the project, follow these steps:

1. Clone the repository:

git clone https://github.com/kushagraghosh/HTTPLoadTesting.git 

2. Go to the project directory:

cd HTTPLoadTesting

3. Build the Docker image:

docker build -t http_load_tester .

Once the Docker image is built, you can run the load testing tool by executing the Docker container with appropriate arguments.

Usage

Command-line Arguments

• URL: The URL of the endpoint to test.

Optional Arguments:

• --num-requests NUM_REQUESTS: Number of requests to send. Default is 10.
• --qps QPS: Queries per second (QPS). Default is 1.0.
• --test-type {load,stress}: Type of test to run. Default is 'load'.

Running Load Test

To run a load test, execute the following two commands:

docker run -it <image_name> /bin/bash

python load_tester.py <URL> [--num-requests NUM_REQUESTS] [--qps QPS] [--test-type {load,stress}]

Example:

docker run -it http_load_tester /bin/bash

python load_tester.py https://www.google.com/ --num-requests 4 --qps 2.0

This command will send 4 requests to https://www.google.com/ with a rate of 2 queries per second.

Running Stress Test

To run a stress test, specify the --test-type stress argument:

docker run -it <image_name> /bin/bash

python load_tester.py <URL> --test-type stress [--num-requests NUM_REQUESTS] [--qps QPS]

Example:

docker run -it http_load_tester /bin/bash

python load_tester.py https://fireworks.ai/ --test-type stress --num-requests 20 --qps 2.0

This command will continuously increase the number of requests sent to https://fireworks.ai/ until system failure or 20, aiming for a QPS of 2.0.