Embodiments are described for an autonomously and dynamically allocating resources in a distributed network based on forecasted a-priori CPU resource utilization, rather than a manual throttle setting. A multivariate (CPU idle %, disk I/O, network and memory) rather than single variable approach for Probabilistic Weighted Fuzzy Time Series (PWFTS) is used for forecasting compute resources. The dynamic throttling is combined with an adaptive compute change rate detection and correction. A single spike detection and removal mechanism is used to prevent the application of too many frequent throttling changes. Such a method can be implemented for several use cases including, but not limited to: cloud data migration, replication to a storage server, system upgrades, bandwidth throttling in storage networks, and garbage collection.

A system for forwarding packets between a first endpoint and a second endpoint, comprising one or more processors; a first network interface for communication with the first endpoint and a second network interface for communication with the second endpoint; and non-transitory memory comprising instructions. The instructions cause the one or more processors to receive a first packet from the first endpoint comprising a first data payload; generate a second packet, comprising the first data payload and an indicator of remaining buffer capacity different from an actual buffer capacity of the system; transmit the second packet to the second endpoint; receive a third packet from the second endpoint comprising a second data payload; generate a fourth packet, comprising the second data payload and an indicator of remaining buffer capacity different from an actual buffer capacity of the system; and transmit the fourth packet to the first endpoint.

Aspects of the present disclosure involve systems, methods, computer program products, and the like, for controlling a congestion window (CWND) value of a communication session of a CDN. In particular, a content server may analyze a request to determine or receive an indication of the type of content being requested. The content server may then set the initial CWND based on the type of content being requested. For example, the content server may set a relatively high CWND value for requested content that is not particularly large, such as image files or text, so that the data of the content is received at the client device quickly. For larger files or files that a have a determined smaller urgency, the initial CWND may be set at a lower value to ensure that providing the data of the content does not congest the link between the devices.

A data collection unit of an NW controller (6) collects, from an L2SW (5) constituting an L2NW (4), uplink observation data indicating traffic volume of a session request packet addressed to an application server 2 and downlink observation data indicating traffic volume of a session response packet transmitted from the application server 2, which are acquired by observing packets input to the L2SW (5). A control unit changes shaping rate of the packets addressed to the application server (2) through the L2SW (5) included in the L2NW (4) based on a ratio between the traffic volume indicated by the uplink observation data and the traffic volume indicated by the downlink observation data, or a ratio between an increment of the traffic volume acquired from the uplink observation data and an increment of the traffic volume acquired from the downlink observation data.

Systems and methods for controlling congestion of a data network are provided. An engine round-trip time (RTT) and a fabric RTT for a network flow are determined. An engine-based congestion window size for the flow is determined based on the engine RTT and a target engine RTT. A fabric-based congestion window size for the flow is determined based on the fabric RTT and a target fabric RTT. The smaller of the engine-based congestion window size and the fabric-based window size is selected for use in transmitting a future packet associated with the flow. The target engine RTT is determined based in part on the current congestion window used to transmit packets for the flow and/or the target fabric RTT is determined based on a number of hops packets associated with the flow traverse from a source to a destination associated with the flow.

A congestion control processing method uses a two-level scheduling manner of a forwarding device and a destination device, where the network device of a data center network performs coarse-grained bandwidth allocation based on weights of flows destined for different destination devices. The network device allocates each flow a bandwidth that does not cause congestion, and notifies the destination device. The destination device performs fine-grained division, determines a maximum sending rate for each flow, and notifies a packet sending device of the maximum sending rate.

Embodiments of this application provide a method, an apparatus, and a system for managing a network slice instance. The method includes: receiving, by a first network device, a network slice instance creation request from a transmit end device, to request to create a target network slice instance; sending, by the first network device, network function configuration indication information to a second network device based on the description information, to instruct the second network device to configure a network function of the target network slice instance; receiving, by the first network device, network function configuration response information sent by the second network device, to indicate that the configuration of the network function of the target network slice instance is completed; and sending, by the first network device, network slice instance creation response information to the transmit end device.

The disclosure relates to an Edge Cloud Platform (ECP) and a method executed in the ECP, for providing a low-latency, distributed, multi-user application. The method comprises determining a first location of a first group of users requesting access to the multi-user application and deploying the multi-user application in a first Point of Presence (PoP) in a first Service Provider (SP) domain operative to serve the first group of users. The method comprises determining a second location of a second group of users requesting access to the multi-user application and deploying a proxy of the multi-user application in a second PoP in a second SP domain operative to serve the second group of users. The method comprises, upon determining that a Software License Agreement (SLA) exists between the first and second SPs, establishing a tunnel for linking the multi-user application and the proxy of the multi-user application, thereby providing the low-latency.

Embodiments are described for an autonomously and dynamically allocating resources in a distributed network based on forecasted a-priori CPU resource utilization, rather than a manual throttle setting. A multivariate (CPU idle %, disk I/O, network and memory) rather than single variable approach for Probabilistic Weighted Fuzzy Time Series (PWFTS) is used for forecasting compute resources. The dynamic throttling is combined with an adaptive compute change rate detection and correction. A single spike detection and removal mechanism is used to prevent the application of too many frequent throttling changes. Such a method can be implemented for several use cases including, but not limited to: cloud data migration, replication to a storage server, system upgrades, bandwidth throttling in storage networks, and garbage collection.

A first network device sends a border gateway protocol (BGP) message to a second network device. The BGP message includes Ethernet virtual private network network layer reachability information (EVPN NLRI) and first information. The first information indicates a first maximum rate at which the second network device sends traffic to the first network device based on the EVPN NLRI. When the second network device sends the traffic to the first network device based on the EVPN NLRI, a first-type rate is less than or equal to the first maximum rate. According to the method in this application, less public network bandwidth is occupied, and overheads of the network device can be reduced.