I don't understand when people go crazy with their servers after getting mentioned on the social web because of the load. It's short lived, no need to panic, just make the page static.
that's one elegant and simple solution, I agree :-) But in this case the load wasn't so targeted, it was spread on 3-4 pages. But you're right, I could have made them all 3 static.
I gotta disagree with the whole "make sure you're ready" idea here. It goes against the launch-and-iterate mantra that I live by.
Always remember, the vast majority of sites don't scale...because they don't have to. Just don't do anything really stupid when you're setting things up at first, and be ready to make changes when (more like if) you ever get popular.
Still, I do agree with not shooting from the hip in the middle of a huge traffic load. Installing new plugins or tweaking code at these times is an awful idea. Much better to call your host and ask to be bumped to a beefier machine. Throw money at the problem and hope it helps.
Gah, wordpress. I serve static pages off of a PII-500mhz with 256MB RAM, and 50 concurrent users look about the same as 1. MySQL is not your friend. I'm using blogger specifically because it pushes static files, but I want to check out http://github.com/mojombo/jekyll (Ruby) or http://github.com/lakshmivyas/hyde (Python).
how does one test this? write a script that uses threads to access pages in parallel? see what happens...temporarily pay for more hardware (i have a virtual machine with linode), gain some understanding, ..., profit
any particular services, snippets or frameworks people use to test this?
For more realistic mutli-page tests Selenium works well - spin up a few EC2 or similar machines, run Selenium Grid and you can get a decent idea of how your site will cope.
For a simple Wordpress blog Apachebench should suffice though.