Route information identifying servers that an e-mail message travels through, along with per-hop latency information, is aggregated. A choke point or other latency-related issue is identified and a control signal is generated to control the e-mail system based upon the identified choke point or other latency-related issue.

A method for routing data packets to a distribution server, comprising generating server load data at the server using a processor. Compiling the server load data into a data update using the processor. Transmitting the data update from the server to one or more routers using a network data transmission system. Modifying a routing algorithm at the one or more routers to utilize the data update using an associated router processor.

A Software Defined Network (SDN) includes a plurality of forwarding devices and a routing control device located separate from the forwarding devices. The routing control device, establishes paths to and from the network forwarding devices. Using such paths, forwarding devices send the routing control device information reflecting the topology if the network. Embodiments disclosed herein enable automatic discovery of the topology of the network and the paths to and from the routing control device.

Conventional internet routing is handled using routing protocols such as the Border Gateway Protocol (BGP). However, simple BGP does not account for latency, packet loss, or cost. To address this problem, smart routing systems that route traffic fast and in a cost-effective manner are implemented. In one approach, smart routing systems measure, compare, and analyze round-trip latencies and other metrics between a customer premises and one or more endpoints. Optimal inbound and outbound transit providers are selected for each endpoint based on these measurements. Other smart routing systems collect and analyze Real User Monitoring (RUM) data to predict latency performance of different content origins for serving data to a particular client based on the client's IP address and the content origins' IP addresses, which are ranked by performance. These rankings are used to steer traffic along lower latency paths by resolving Domain Name System (DNS) queries based on the performance associated with the IP addresses.

A network includes a plurality of endpoint routers and intermediate routers. When a new data stream is detected at any endpoint router, the first packet is sent to a virtual routing server with knowledge of the entire network topology. Based on the topology, current usage, and historical usage, the virtual routing server determined a path for the data stream and begins to update the routing tables of the intermediate routers to reflect the determined path. Until the update is complete, all packets in the data stream are routed first to the virtual routing server and then to their destination. Once the update is complete, packets in the data stream are routed directly along the determined path.

A network service allocating system for a software defined network (SDN) includes an SDN controller, a plurality of SDN switches, a cloud server, and a local server. The SDN controller includes modules for service managing and path managing. Each SDN switch can receive a packet from a client and send the packet to the SDN controller. The service managing module analyzes the packet to identify type of service required and allocates the network service to the cloud server or to the local server according to the type of service required. The path managing module plans an optimum transmission path and sends the path to the SDN controller. Thereby, the SDN switch can obtain the network service. An SDN controller and a network service allocating method are also provided.

Interactive-streaming-based applications, such as cloud gaming, giga-pixel streaming and virtual reality, have rigorous requirements on the network latency, which can be satisfied by routing users' requests over an overlay network. Existing overlay routing strategies suffer from high deployment and maintenance costs. An optimized cloud overlay routing system and method is discussed herein, which maximizes the number of user requests for the interactive-streaming-based applications to be served, lower the deployment and maintenance costs for the overlay services, reduce the overall network delay, and balance the network loads by bypassing busy underlay links.

In one embodiment, a multicast listener device floods a path lookup request to search for a multicast tree, and may then receive path lookup responses from candidate nodes on the multicast tree, where each of the path lookup responses indicates a unicast routing cost from a respective candidate node to the multicast listener device, and where each of the candidate nodes is configured to suppress a path lookup response if a total path latency from a source of the multicast tree to the multicast listener device via that respective candidate node is greater than a maximum allowable path latency. The multicast listener device may then select a particular candidate node as a join point for the multicast tree based on the particular node having a lowest associated unicast routing cost to the multicast listener device from among the candidate nodes, and joins the multicast tree at the selected join point.

A communication path management apparatus includes a memory, and a processor coupled to the memory and to decide, for each of processing-pairs, part of inter-adjacent-server communication paths that couple adjacent servers on communication paths among servers in the network as data-transfer-paths used for data transfer from a first processing to a second processing of a processing-pair, the processing-pair including first and second processings that are consecutive in the order of execution, based on an execution track record of the first processing that is earlier in the order of execution and the second processing that is later in the order of execution in the processing-pair in each of the servers, and instruct servers at both ends of the data-transfer-paths decided regarding each of the processing-pairs to carry out data transfer of an execution result of the first processing by the data-transfer-paths.

A route determining method and apparatus, and a communications device are provided. The method includes: obtaining startpoint information and endpoint information of a required path, where the startpoint information includes a sequence number of a startpoint POD and a sequence number of a startpoint edge switch, and the endpoint information includes a sequence number of an endpoint POD and a sequence number of an endpoint edge switch; determining, from an n-dimensional Latin square, an element whose row is the sequence number of the startpoint edge switch and column is the sequence number of the endpoint edge switch; and if the sequence number of the startpoint POD is the same as the sequence number of the endpoint POD, determining that the required path is the startpoint edge switch, an aggregation switch corresponding to the element in the POD, and the endpoint edge switch.