What was Facebook's early architecture?

Facebook's early architecture faced significant challenges and evolved through necessity. Initially, the architecture involved a simplistic setup with servers named after universities like Stanford and Harvard. Each server handled a separate PHP instance for different colleges, meaning the code was copied and pasted across servers to scale the platform 1.

As Facebook grew, it transitioned into a client-server architecture with multiple servers handling different tasks. For instance, Newsfeed ranking was processed on special servers that communicated with the main web server through a remote procedure call (RPC) system. The initial RPC system was unreliable, leading to the development of a more robust system called Thrift 2.

These architectural choices highlight Facebook's early focus on quickly scaling and adapting to user needs, even if it meant dealing with technical and logistical challenges along the way.