After learning how to load balance, you still need to keep your web files consistent between your web servers. My tool of choice for doing so is rsync which includes smart features such as delta uploads (if it notices a file has changed it will only upload the difference, not the whole file from scratch).
I am assuming that you will have a particular “main” web server which you always update with new content first. The new content can either be “pushed” by the main web server to dependent web servers running rsync daemons, or it can be “pulled” by dependent web servers from the main web server. I suggest running a “pull” environment because your main web server will not need any knowledge of the existence of the dependent web servers.
On a recent multi person project, we’ve used a subversion client to directly pull the latest project files into the web directory. We do so because its a complicated environment that we have not yet created individual sandboxes for. To test a change, the code must be committed to subversion and then we execute “svn export” on the webserver to pull the latest files from our subversion repository directly to the web directory. The only downside seemed like we were going to have a crazy number of revisions. But we’ve also run into one other problem: APC.
