A couple of days ago, I needed to measure the performance of a category page in a Magento shop, with different combinations of sorting and filtering parameters. I needed the test to be objective and reproducible, in order to check for performance before and after some changes.

Well, this sounds like a perfect job for a cURL + a bash script! I have written a short one1, which just takes an input url and appends the parameter no_cache=1 afterwards, in order to measure the load time with and without cache. Here’s the output:

$ ./benchmark-url.sh http://www.example.com/
Checking response time for http://www.example.com/ (cached) - 0,189899
Checking response time for http://www.example.com/?no_cache=1 (not cached) - 0,207218

The script leverages a great feature of cURL2, which can also return additional information about the transaction: for instance, the duration (with the parameter -w or --write-out and a formatting string):

curl -s -w %{time_total}\\n -o /dev/null "www.example.com"

In this example, I was just looking at the total time of the operation, but the report can become very granular if you also check for other values, e.g. lookup time, connect time, and many more metrics3.

Notes

  1. Check out the script on GitHub ↩

  2. cURL Basic Tutorial for HTTP Timing Measurement - Medium ↩

  3. A more detailed description of cURL and its capabilities - HTTP Transaction Timing Breakdown with cURL ↩