I can't speak for him, but the RSS reader I am working on at the moment is similarly split into multiple parts:
* Crawler - retrieves RSS feeds and parses articles
* Database server
* Analyser - performs semantic analysis and various other statistical processing on posts
* Propagator - pushes new posts to everyone's feed
* Cacher - serves a cached copy of the user's feed
* Web server
Based on loads I would just add more nodes depending on what was needed.