This authors’ definition of a server and answer to the rethorical question in the title.
The word server is derived from the verb “to serve” and the purpose of the server is to serve clients in the network. The word is used to describe two different interpretation on is the actual computer hardware and the other is the application running on the computer. The same computer hardware can both act as a server and a client. The term “dedicated server” refers to a physical computer that only acts as a server, and is often only used to supply one service.
A server is an application that runs on a computer. The application supplies some form of service to clients over the network. The serves purpose is to share a common resource on the network to the clients, e.g. access to database or managing print queues for network printers. The server service is implemented on the application layers of the OSI reference model, layer 5-7, the session layer, the presentation layer and the application layer. A server is not proactive, but is passive and waits for input from the clients to take some form of action. That is, that the server listens on a socket specified by a port number, and responds to the client or take some form of action based on the programming of the server application.
For a server to handle multiple connections, or more than one client at the same time, the server application must be threaded. That means that the application must run several instances of the services implementation. The server starts a new instance of the thread when a client connects to the server, to solely handle the communication with that specific client.
A general description of servers.
The official OSI reference model at iso.org (You must enable cookies)
The OSI reference model at en.wikipedia.org
Article about threading at en.wikipedia.org