Lately, I’ve been tuning a Quarkus backend and got curious: How many concurrent requests can this thing actually handle? That’s when I stumbled upon wrk, a modern HTTP benchmarking tool that’s fast and flexible — but also a little unfriendly if you’re just starting out.

So I built a wrapper: wrknoob — a simple, noob-friendly Python CLI that wraps around wrk, runs structured load tests, and shows beautiful terminal tables and plots. Here’s how it went.

The Problem

I could run commands like:

1wrk -t8 -c100 -d15s http://localhost:8080/hello

But over time, I wanted:

• To test across multiple concurrency levels
• To automatically visualize requests/sec and latency
• To save results to CSV
• To avoid flipping between notes and terminal

I didn’t want a full-fledged Grafana dashboard or a k6 setup — just something quick, scriptable, and pretty.

Enter wrknoob

I created a Python CLI tool that:

✅ Asks for input parameters interactively
✅ Runs wrk for each concurrency level
✅ Parses and displays the output beautifully with rich
✅ Plots graphs using matplotlib
✅ Can save results to CSV and images

Features in a Nutshell

• 🎛️ Prompt-based test configuration
• 📊 Graph of requests/sec across concurrencies
• 📉 Latency trend analysis
• 🧾 Rich terminal table for results
• 💾 Export to report.csv and plot.png

Screenshot

Above: Results table in terminal and corresponding plot

Installation

You’ll need Python 3.8+ and wrk installed. On macOS:

1brew install wrk

Then:

1git clone https://github.com/sitasp/wrknoob.git
2cd wrknoob
3python3 -m venv .venv
4source .venv/bin/activate
5pip install -r requirements.txt

Run with:

1python wrknoob.py

Sample Run

1Target URL: http://localhost:8080/hello
2Test Duration (e.g., 10s): 10
3Threads: 4
4Concurrent Connections (comma-separated): 10,25,50,100
5Show Graph? [y/n]: y
6Save Report to CSV? [y/n]: y

Output:

• report.csv — with columns like concurrency, req/sec, latency
• plot.png — a visual trend of performance

Why I Built This

Load testing felt like a mysterious black box — but writing wrknoob helped me understand:

• The relationship between latency and concurrency
• How saturation begins to degrade performance
• That high req/sec != efficient if latency spikes

Final Thoughts

If you’re testing your APIs, especially on localhost, wrknoob helps you benchmark easier without sweating over wrk flags. It’s not meant for enterprise-scale simulation, but it’s perfect for developers experimenting with performance.

Check out the repo here, and feel free to open issues or PRs!

Happy benchmarking!