| 摘要 |
An improved TCP/IP protocol for use in a client/server network is disclosed. The improved protocol allows a server to switch execution of an application from one processor or task to another processor or task without interruption of the IP connection or serious delay in execution of the application by a server or client. As implemented, each server usually includes a unique hardware address, an improved IP layer with a shared IP address, a "Speak No Evil-Communication Protocol" layer with a unique SNE-CP address, a soft switch protocol ("SSP") layer with at least one unique SSP address assigned to a process or task, usually a TCP layer, at least one SSP monitoring process and a processor executing at least one software application. In this invention, each server typically receives each message packet from each client; however, only one server is allowed to respond to the client's message. Thus, in the event that one processor or task faults, the application can be switched to another processor or task without interruption of the IP connection or serious delay in execution.
|
