I have an html directory filled with very small PHP scripts that do nothing but transform query string parameters to a redirect.
The apache server, which takes ~60K requests per day is not configured with any fast cgi/mod cgi (i’m not sure the correct term here), and it’s not a particularly beefy machine, so I hesitate to add significant overhead to the logic in these scripts.
What I must do is introduce a lookup into an associative array or equivalent so a string in the inbound URL can be mapped to a number (a string really) on the outbound url.
My fear is that if I introduce a MySQL query on every request, I’ll kill the performance.
Likewise, if I include a PHP file containing the definition of an array with 1000 items, I fear the lack of any memory resident PHP interpreter will cause a similar amount of overhead.
I had the thought to write a shell script and leverage the _ENV global, but I wonder if that’s going to be just as much overhead.
Any thoughts from a more experienced LAMP developer would be greatly appreciated.
Possible answers:
- tell me that configuring an apache integrated PHP is nothing to fear and give me a link to some instructions
- tell me that a MySQL query for each request is negligible overhead
- tell me that including a large array definition on reach request is negligible overhead
- …some other idea…
What if we told you to “test it and see”?
If you have memory, MySQL should be negligible. It’s a socket hit to a memory cache of a simply query.
If you have CPU, the PHP module will likely not be noticeable.
If you have neither, you’re doomed either way :).
60K requests per day, over an 8 hour day, is 2 transactions per second. That’s really not a lot of traffic. No doubt you have spikes and heavier and quieter times. But even still.
Embedding PHP in to apache will, in general, speed things up. But not by as much as you think. The cost of the startup is the fork and creation of the process, but the PHP interpreter code is already loaded in to RAM somewhere else (likely if you have requests overlapping), or minimally in disk cache. So it will be measurable, can’t say it will be noticeable. (Sprinkle “it depends” like fairy dust all over all of these statements, right?)
Simply put, if you can test it, I’d test the 1000 lines in the php map and see how it goes. For a simple example, just time it from the command line (several times) and see what the difference is. This file will likely stay hot in the file cache, so I/O will be minimal, thus putting the bulk of the burden on the PHP interpreters parser (and thus CPU).
You say you don’t have a beefy machine, so I don’t know what that means, but modern “slow” machines are “pretty beefy(tm)”.