http://aosabook.org/en/index.html
A good example is Scalable Web Architecture and Distributed Systems by Kate Matsudaira:
http://aosabook.org/en/distsys.html
https://meta.wikimedia.org/wiki/Wikimedia_servers
At what level of scale might one expect to need what's going on in the "Edge Cluster", as opposed letting all the requests fly right into the app servers?
http://aosabook.org/en/index.html
A good example is Scalable Web Architecture and Distributed Systems by Kate Matsudaira:
http://aosabook.org/en/distsys.html