Europe is a system to allow for the distribution of CPU cycles for user-space applications. Traditionally distributed computing systems are focused on sheer computing power and don’t worry about the things computer users worry about because of the sheer size of the calculations being preformed. User-space applications, on the other hand, have to worry about the user. Sheer FLOPS wont cut it in user-space because the task needs to be completed in real-time. User perception, timely task completion, and network latency become an issue right away.
Server : Hardware dedicated to providing CPU cycles for the network.
Network Resources : Computing power, or CPU cycles, that is available for use on the network.
Client : A user system that uses positive network resources and provides tasks for the network to preform.
Super-client : A user system that uses negative network resources and may provide CPU cycles for network use.
- Operations are time-constrained. Tasks need to be completed at speeds similar to local operations, network wide.
- Network latency must be taken into account. See no. 1.
- Geographically oriented: Local servers first, followed by nearby super-clients, followed by the local machine. See no. 2.
Europe creates a hierarchical network of inter-connected servers, each of which are connected to geographically similar super-clients, each of which manage a cluster of clients. Consequently a client will only make intermittent contact with the server while working closely with s super-client that is closer geographically to the client system.