|
Why not do a mix of both? Instead of rebuilding the pages every X minutes - which itself will cause periodical CPU spikes - cache them. If the page requested hasn't yet been generated, generate it; if it has, then just readfile() the cache'd page. This should spread your load out nicely since most of the time you'll just be serving out a static page.
(I would personally use a 404 handler which creates true static .html if the page doesn't already exist, but that gets a bit more complicated config wise...)
|