Multiple processors are often used in computing systems to solve very large, complex problems, such as those encountered in artificial intelligence. Such processors typically exchange data among each other via an interconnect fabric (such as, e.g., a group of network connections and switches) in solving such complex problems. The amount of data injected into the interconnect fabric by the processors can at times overwhelm the interconnect fabric preventing some of the processors from communicating with each other. To address this problem, techniques are disclosed to enable, for example, processors that are connected to an interconnect fabric to coordinate and control the amount of data injected so that the interconnect fabric does not get overwhelmed.

Systems and methods are provided for operating a media transmission network. The system includes at least one destination device for receiving a plurality of media streams from a plurality of source devices. The system further includes a controller that is configured to, for each media stream of the plurality of media streams: determine a media property adjustment for the media stream based at least on the media stream; identify a source device from the plurality of source devices associated with generating the media stream; determine at least one device setting for the identified source device to apply the media property adjustment to the media stream; generate a control packet for configuring the identified source device based on the at least one device setting, the control packet including the at least one device setting; and transmit the control packet to the identified source device.

Techniques for dropping packets at congested network elements for no drop traffic are described. A network element in communication with a congested network element initiates a copy packet queue and stores a copy of each transmitted no-drop packet sent to the congested element. When the network element receives an indication that the congested element has dropped a no-drop packet, the network element begins retransmission of the dropped packets to the congested element from the copy packet queue, thus providing a lossless network while allowing for dropped packets.

Systems and methods for limiting calls to access a cloud-based system are disclosed. The systems and methods obtain a rate limiting policy including at least one attribute and a counting interval, the at least one attribute including at least one of a username associated with a client, an instance, an organization associated with the client, a resource being requested, a service being requested, a geographical access region, and an Application Programming Interface (API) being requested. The systems and methods also mark an entry, based on the rate limiting policy, in a database for each call the client makes. The systems and methods further enforce the rate liming policy by not processing calls from the client associated with the at least one attribute that are made for a count of calls marked that is beyond the counting interval.

A network element includes at least one headroom buffer, and flow-control circuitry. The headroom buffer is configured for receiving and storing packets from a peer network element having at least two data sources, each headroom buffer serving multiple packets. The flow-control circuitry is configured to quantify a congestion severity measure, and, in response to detecting a congestion in the headroom buffer, to send to the peer network element pause-request signaling that instructs the peer network element to stop transmitting packets that (i) are associated with the congested headroom buffer and (ii) have priorities that are selected based on the congestion severity measure.

In some examples, a non-transitory machine-readable medium can include instructions executable by a processing resource to: monitor a quantity of interactions between a plurality of user interfaces with an embedded device, determine when the quantity of interactions with the embedded device exceeds a threshold, send a slow-down message to a portion of the plurality of user interfaces in response to the interactions with the embedded device exceeding the threshold, and restrict a portion of the quantity of interactions with the embedded device when the quantity of interactions continues to exceed the threshold for a quantity of time after the slow-down message was sent to the portion of the plurality of user interfaces.

A network element includes multiple output ports and circuitry. The multiple output ports are configured to transmit packets over multiple respective network links of a communication network. The circuitry is configured to receive from the communication network, via one or more input ports of the network element, packets that are destined for transmission via the multiple output ports, to monitor multiple data-counts, each data-count corresponding to a respective output port, and is indicative of a respective data volume of the packets forwarded for transmission via the respective output port, to select for a given packet, based on the data-counts, an output port among the multiple output ports, and to forward the given packet for transmission via the selected output port.

A method may include receiving a first message requesting a change associated with an Ethernet service that includes routing data from a user device to a destination device via a first service provider and a second service provider. The method may also include determining, by the first service provider, whether resources are available in a first network associated with the first service provider to fulfill the requested change and determining, by the second service provider, whether resources are available in a second network associated with the second service provider to fulfill the requested change. The method may further include sending a second indicating whether the change is accepted and in response to determining that the change is accepted, automatically implementing the change for the Ethernet service.

One example method includes creating a manifest that specifies one or more requirements concerning execution of an application that resides at an end device in an N-tier configuration, identifying a workload that is associated with the application and executable at one or more edge stations of the N-tier configuration, gathering and evaluating network telemetry, orchestrating the workload based on the network telemetry and the manifest, scheduling performance of the workload at the one or more edge stations, and performing the workload at the one or more edge stations in accordance with the scheduling.

Techniques and architectures for measuring available bandwidth. A train of probe packets is received from a remote electronic device. A network transmission delay for at least two packets from the train of probe packets is measured. Network congestion is estimated utilizing the at least two packets from the train of probe packets. An estimated available bandwidth is computed based on the network transmission and estimated network congestion. One or more network transmission characteristics are modified based on the estimated available bandwidth.