| 摘要 |
The arrangement described relates to routing of packets and providing load-balancing across a network. A relay system has a router 24 and multiple relay servers 30, 32, 34 with specific network addresses on a first network 2, e.g. the internet. The router and servers are mutually connected by a second network 16, e.g. a private access controlled network. The router receives a relay-service request from a source 12 on the first network, addressed to an address distinct from the relay servers' specific addresses. It forwards the request to the relay servers, possibly using an e.g. Ethernet broadcast address (common destination address). Each relay server that receives the request instructs the source to re-issue or re-direct a relay-service request to its specific address, e.g. specific IP address. The server that receives the relay-service request provides the service. The relay servers may be Traversal Using Relays Around NAT (TURN) servers or STUN servers. The arrangement described achieves a load-balancing mechanism across the relay servers as the relay server that responds to the request from the source first will generally be the lightest loaded server which then directs the source to re-direct or re-issue the request directly to itself. The arrangement described may be applied to e.g. a video call between a source 12 and destination 14 such that Quality of Service (QoS) guaranteed links are utilized (in the private network 16) and time for traffic between the source and destination in non-guaranteed QoS links (e.g. in the internet 2) is minimized. |
