Ok... I have some more test results. After optimizing as much as I can
think of, without using caching, I've gotten things down to a 13x
difference. Using Apache's ab performance test, the image comes from a file
at an average of 2ms and from the database (using PHP4) at an average of
I know... it just reiterates what you were already saying, but it sure is
great to see actual numbers measuring the difference. Maybe the difference
could be even less if I were properly optimizing MySQL.
The big question still outstanding, for me at least, is whether web page
caching makes the performance difference a mute point. If caching is
storing everything as files, we get the best of both worlds.
Plus, I think there may be a little bit of a security benefit. A directory
has to be marked as writeable so that scripts can store image files. This
isn't necessary when using MySQL.
Do you agree with the security benefit? Does webpage caching negate the
> -----Original Message-----
> Grabbing the file was 38 times faster because MySQL was not designed
> to be a filesystem. There are filesystems out there specifically
> designed to handle hundreds of thousands of small files. One of the
> best is ReiserFS http://www.namesys.com
> If you record the filename in mysql tracking becomes a non issue.