| 摘要 |
A data communications network having a number of servers and a number of service providers employs an improved protocol for translating service names to network addresses. A server sends a multi-cast message to all service providers requesting a service needed by one of the user terminals connected to this server. In the request, the service is identified by function, rather than by address. Any node which provides this service, receiving the request and able to reply, sends a reply message immediately to the server. The reply message identifies the node (service provider) by network address, so a transaction with this provider can be initiated by the server immediately. If the service provider cannot reply, a third party node can send a reply to identify the provider by address. If the server does not receive a reply from either the provider or a third party, then a cache of network addresses of services most recently used by the server is searched for this particular service. The cache may include a separate local cache for each user terminal connected to this server, with each local cache containing the service and address for the last-used service for one of the user terminals. In addition, a secondary cache is maintained for all items replaced in the local caches when they are updated. When the cache is searched (both local and secondary), if the service address is found, a transaction is begun immediately by the server to utilize the service. If not found in the cache, the server waits for a service advertisement to appear on the network, and obtains the network address of the provider from the advertisement.
|
