The HTTPD documentation says "You should avoid using .htaccess files completely if you have access to httpd main server config file. Using .htaccess files slows down your Apache http server. Any directive that you can include in a .htaccess file is better set in a Directory block, as it will have the same effect with better performance."
I wanted to see if I could measure how much slower it is and boy does it make a difference. (This is actually part of my PhD research into how to better understand configuration management.)
I built HTTPD from source with a lot of debugging features (i.e. symbols, no compiler optimization), so these specific numbers are only valid in the context of this test, but they are still interesting.
I created a file DOCUMENT_ROOT/1.txt containing just the text "1" (Short URL) and a file DOCUMENT_ROOT/1/2/3/4/5/6/7/8/9/10/11/12/13/14/15/16/17/18/19/20/21.txt containing the text "21" (Long URL).
I used ab to run get requests, first 10,000, then 100,000, then 100,000 again, just to check for variability (and it turns out there isn't any significant variability). I ran ab on the same machine as httpd.
I ran the requests first with AllowOverride None, then with AllowOverride All but no .htaccess files, then with AllowOverride All and .htaccess files with a mix of "Require all denied" and "Require all granted".
I also collected the number of instructions executed by the system using perf.
Here is a quick summary of the results:
AllowOverride None / Long URL:
- 1.367 seconds
- 13.637 seconds
- 13.607\ seconds
AllowOverride None / Short URL:
- 1.283 seconds
- 12.981 seconds
- 12.989 seconds
AllowOverride All / Long URL:
- 2.002 seconds
- 20.015 seconds
- 20.032 seconds
AllowOverride All / Short URL:
- 1.370 seconds
- 13.581 seconds
- 13.590 seconds
AllowOverride All / Long URL with `.htaccess` files:
- 3.062 seconds
- 31.042 seconds
- 31.122 seconds
AllowOverride All / Short URL with `.htaccess` files
- 1.431 seconds
- 14.487 seconds
- 14.461 seconds
The change in perf counters matched the changes in wall clock time.
The only thing surprising about any of these results was the magnitude of the performance effects.
I think it is most interesting that for this example of a path 20 directories deep, having an extra .htaccess file nested in each directory actually doubled the amount of time it took to process the request.
- Y
