It is difficult to give an exact answer without knowing which technologies you plan to use. The approach to the application may not be completely unaware of the planned physical infrastructure, if scaling is the main driver.
Caching should be a big problem. There are also ways to expand the equipment in which your data lives.
A very interesting and instructive reading is a real biography of a live magazine, a history of scaling , and how they grew up in physical presence with massive growth on their website. One of the main responses from their work was the new caching technology, memcached, which is now used by FaceBook and others. This is surprisingly honest.
source share