Client/Server Telephony
In a client/server environment, TAPI provides distributed access to shared telephony resources. For example, a LAN-based server might have multiple telephone-line connections to a local telephone switch or PBX. TAPI operations invoked at any associated client are forwarded over the LAN to the server. The server uses third-party call control between the server and the PBX to implement the client's call-control requests.
This model offers a lower cost per computer for call control if the LAN is already in use.
The server can be connected to the switch using a switch-to-host link. It is also possible for a PBX to be directly connected to the LAN on which the server and associated clients reside. Within these distributed configurations, different subconfigurations are possible:
To provide personal telephony to each desktop, the service provider could model the PBX line associated with the computer (on a desktop) as a single line device with one channel. Each client computer would have one line device available.
Each third-party station can be modeled as a separate line device to allow applications to control calls on other stations. (In a PBX, a station is anything to which a wire leads from the PBX.) This enables the application to control calls on other stations. This solution requires that the application open each line it wants to manipulate or monitor.
The set of all third-party stations can be modeled as a single line device with one address (one phone number) assigned to it per station. Only a single device is to be opened, providing monitoring and control of all addresses (all stations) on the line.
A major advantage of such client/server implementations is a lowered cost of telephony services per client application.