How to Benchmark Web Servers with http_load, webbench, ab, and Siege

1. http_load

http_load is a lightweight Linux performance testing tool (≈100 KB) that runs in a single process, can test both HTTP and HTTPS, and measures web‑server throughput and load without overwhelming the client.

wget http://www.acme.com/software/http_load/http_load-12mar2006.tar.gz
tar zxvf http_load-12mar2006.tar.gz
cd http_load-12mar2006
make && make install

Command format:

http_load -parallel <processes> -seconds <time> urls

. Parameters can be combined freely: -parallel (or -p): number of concurrent user processes -fetches (or -f): total number of requests -rate (or -r): requests per second -seconds (or -s): total test duration urls: target URL or file containing many URLs (50‑100+ recommended)

Example output (excerpt):

#http_load -rate 2 -seconds 300 urls
591 fetches, 8 max parallel, 5.33606e+06 bytes, in 300 seconds
# ... (additional metrics) ...
HTTP response codes: code 200 -- 591

Key metrics to consider are fetches/sec and msecs/connect; also monitor CPU, memory, etc., for a comprehensive performance assessment.

2. webbench

webbench is a Linux web‑server stress test tool capable of simulating up to 30 000 concurrent connections.

wget http://tmublr.com/_media/blog/webbench-1.5.tar.gz
tar zxvf webbench-1.5.tar.gz
cd webbench-1.5
make && make install

Usage: webbench -c <concurrency> -t <seconds> URL.

Example:

#webbench -c 500 -t 30 http://127.0.0.1/test.jpg
Webbench – Simple Web Benchmark 1.5
Benchmarking: GET http://127.0.0.1/phpinfo.php
500 clients, running 30 sec.
Speed=3230 pages/min, 11614212 bytes/sec.
Requests: 1615 succeed, 0 failed.

Analysis: 3230 pages/min ≈ 53.8 pages/sec, and 11.6 MB/sec throughput.

3. ab

ab (ApacheBench) is a powerful testing tool bundled with Apache.

For detailed parameters, see the referenced guide.

4. Siege

Siege is an open‑source stress testing tool designed to evaluate web‑application performance under load. It can simulate multiple users accessing a site, record response times, and repeat tests with configurable concurrency. Unlike ab, Siege can request random URLs from a preset list, making it suitable for realistic load simulation, though ab remains more accurate for pure benchmark tuning.

For more information, refer to the linked detailed guide.