31 results about "Elephant flow" patented technology

Disclosed in the present invention is an SDN-architecture-based routing method for guaranteeing network QoS. The method comprises: on the basis of network topology information, performing abstraction to obtain a directed graph; then, on the basis of an improved K-means algorithm, dividing data into mouse flows, elephant flows and mammoth flows; defining a traffic transmission constraint condition; constructing a DQN neural network; training the DQN neural network by using the maximum Q value, which corresponds to an action a having the maximum reward function R, as an objective, so as to obtain a QoS guarantee routing model; obtaining an optimal path weight value; determining a data packet sending path on the basis a Dijkstra algorithm; and finally issuing path information to a switch. The present invention avoids the situations in which data processing performed by using a conventional K-means algorithm is not stable enough and a classification result is not accurate enough. During the training of a DQN neural network, data is extracted from an experience pool according to the priority of data, such that the neural network pays more attention to important data, thereby improving the learning efficiency of the neural network and the performance of the algorithm.

Some embodiments provide a forwarding element that detects and handles elephant flows. In detecting, the forwarding element of some embodiments monitors statistics or measurements relating to a data flow. In handling, the forwarding element marks each packet associated with a detected elephant flow in some manner to differentiate it from a packet associated with a mouse flow. Alternatively, the forwarding element of break elephant flows into a number mouse flow by facilitating in sending packets associated with the detected elephant flow along different paths.

Provided in the present application are a method and apparatus for solving an IPSec elephant flow on the basis of a DPDK. The method comprises: on the basis of a configuration file, initializing a CPU0 and a CPU1 into PMD thread cores, and initializing CPU2-CPU4 into encryption / decryption engine operation cores, creating lock-free rings for communication between the PMD thread cores and the encryption / decryption engine operation cores, and establishing a network card sending queue and a network card receiving queue; the PMD thread cores regularly executing polling operations, acquiring messages, detecting an elephant flow and a common flow, performing load balancing on the elephant flow, and distributing the messages to the encryption / decryption engine operation cores; and encrypting / decrypting the messages in the encryption / decryption engine operation cores, then sending the messages to the CPU1 for sorting, and then sending the messages to the network card sending queue. In this way, the detection of an elephant flow is realized, and multi-core IPSec encryption / decryption is performed on the elephant flow to solve the performance bottleneck of a single-core processor. The present application can be applied to wider application scenarios.

The invention discloses a method for identifying and processing elephant flow in a distributed energy safety data acquisition device, and relates to the technical field of data flow transmission. Network flow data are acquired through flow acquisition equipment, and data packet characteristic values in unit time are counted through a sliding window; calculating a data flow coefficient according to the data packet characteristic value to judge and generate a first elephant flow processing signal, constructing a CPU utilization rate prediction model based on a convolutional neural network, substituting the real-time CPU utilization rate into the CPU utilization rate prediction model to output an expected CPU utilization rate, and judging to generate a second elephant flow processing signal. The elephant flow is graded through a grading response mechanism to obtain a grading result, a differential processing strategy is executed according to the grading result, the first-level elephant flow adopts multi-channel shunt transmission, the second-level elephant flow starts dynamic encryption transmission, the third-level elephant flow triggers local cache, and a safety early warning signal is generated to execute a safety early warning mechanism.

The embodiment of the invention provides a traffic scheduling method and device, electronic equipment and a medium, and relates to the technical field of communication, and the method comprises the steps: receiving the transmission rates of a plurality of specified data streams reported by a switch; determining the specified data stream of which the transmission rate is greater than a preset rate threshold as an elephant stream; when the outbound port on the switch is congested, selecting an elephant stream to be adjusted from all elephant streams needing to be forwarded by the outbound port; and issuing configuration information of a forwarding strategy of the to-be-adjusted elephant flow to the switch, wherein the forwarding strategy is to forward the message of the to-be-adjusted elephant flow packet by packet through each output port matched with the to-be-adjusted elephant flow. By adopting the method, the network congestion can be reduced, and the bandwidth utilization rate and the forwarding efficiency can be improved.

In some embodiments, a method receives a packet for a flow associated with a workload. Based on an indicator for the flow, the method determines whether the flow corresponds to one of an elephant flow or a mice flow. Only when the flow is determined to correspond to an elephant flow, the method enables a hardware acceleration operation on the packet. The hardware acceleration operation may include hardware operation offload, receive side scaling, and workload migration.

The invention relates to an elephant stream transmission method and device, network equipment, a storage medium and a program product. The method comprises the following steps: identifying quintuple information and queue pair information of an elephant stream to be processed; dividing the elephant flow to be processed into a plurality of sub-workflows according to the current rate of the elephant flow to be processed, the maximum bearing capacity of each network path and the queue pair information; each network path is determined according to the quintuple information; and according to the priority information of each sub-workflow and the delay information of each network path, allocating a corresponding network path to each sub-workflow so as to transmit each sub-workflow. By adopting the method, the problem of elephant flow congestion can be effectively solved.

The invention discloses an ECN-based big and small flow distinguishing scheduling congestion control method and device and a medium, and the method comprises the steps: simulating a flow model of a target network before business deployment, and calculating a first ECN waterline parameter suitable for elephant flow and a second ECN waterline parameter suitable for mouse flow; receiving the service flow forwarded by the network equipment, and detecting the message characteristics of each message in the service flow; the message features comprise a message flow rate, a message size and a message DSCP attribute value; dynamically identifying elephant flow and mouse flow in the service flow based on the message features; and applying the first ECN waterline parameter to a first scheduling queue corresponding to the elephant flow, and applying the second ECN waterline parameter to a second scheduling queue corresponding to the mouse flow, so as to execute differentiated ECN congestion control scheduling on the two queues.

Disclosed in the present invention is an SDN-architecture-based routing method for guaranteeing network QoS. The method comprises: on the basis of network topology information, performing abstraction to obtain a directed graph; then, on the basis of an improved K-means algorithm, dividing data into mouse flows, elephant flows and mammoth flows; defining a traffic transmission constraint condition; constructing a DQN neural network; training the DQN neural network by using the maximum Q value, which corresponds to an action a having the maximum reward function R, as an objective, so as to obtain a QoS guarantee routing model; obtaining an optimal path weight value; determining a data packet sending path on the basis a Dijkstra algorithm; and finally issuing path information to a switch. The present invention avoids the situations in which data processing performed by using a conventional K-means algorithm is not stable enough and a classification result is not accurate enough. During the training of a DQN neural network, data is extracted from an experience pool according to the priority of data, such that the neural network pays more attention to important data, thereby improving the learning efficiency of the neural network and the performance of the algorithm.

The invention discloses a congestion large flow rerouting method and system based on network state perception, and the method specifically comprises the steps: carrying out the statistical calculation of the contribution degree of each flow to a queue in a congestion state, dynamically determining a large flow judgment threshold value through combining with the number of active flows, and recognizing the flow with the contribution degree exceeding the large flow judgment threshold value as a responsibility elephant flow; based on the liability elephant flow list and the network state information, modeling the rerouting problem into a mixed integer programming model with the goal of minimizing the maximum link utilization rate, and solving to obtain a global optimal path allocation scheme; and according to the global optimal path distribution scheme, a new path forwarding rule is installed on each switch and an old path rule is deleted by adopting a first-building and second-dismounting strategy. According to the invention, through cross-layer cooperation of the data plane and the control plane, sub-second congestion response and global optimal flow scheduling are realized, and the problem of congestion migration caused by traditional reactive rerouting is effectively avoided.

The invention relates to a comprehensive path optimization method for a wide-area cloud architecture network in the field of network cloud fusion. The method is particularly applicable to load balancing scheduling of a large number of continuous service flows (elephant flows) in a cloud architecture network. Comprising the following steps: monitoring a network state and identifying elephant flow, collecting information of each link, and detecting and identifying the elephant flow; topological path pre-screening and candidate path library construction improve the calculation and optimization efficiency. Initializing ant colony parameters, and formulating a multi-dimensional heuristic function; performing path exploration and probability selection, and generating an ant colony exploration calculation formula according to a target; dynamically updating pheromones, and searching an optimal path according to local pheromone updating and global pheromone updating; issuing an optimal path and updating a flow table; and continuous monitoring and dynamic adjustment are carried out. According to the invention, a dual control plane combining a routing protocol and an SDN protocol is designed, the improved ant colony algorithm based on network topology is adopted to optimize elephant flow path selection under a cloud network architecture, and the network global service transmission efficiency is improved.

The invention belongs to the field of information communication, and discloses a data center network flow resource allocation method and a related device, and the method comprises the steps: obtaining the state information of each current network flow of a data center; and calling a preset data center resource allocation model according to the state information of each current network flow of the data center to obtain a path of each network flow type of each current communication host pair of the data center and the bandwidth of each hop of the path. The preset data center resource allocation model can dynamically allocate corresponding paths and bandwidths to different network traffic types of each communication host pair of the data center, so that differentiated resource scheduling for dynamically adjusting differentiated demands of different network traffic types, such as mouse flow and elephant flow, is realized. Network congestion caused by disordered resource occupation of elephant flow in a traditional method is avoided, meanwhile, the low-delay requirement of mouse flow is guaranteed, and the transmission efficiency, the network throughput, the link utilization rate and the overall service quality of data center network flow are improved on the whole.

The embodiment of the invention provides a traffic scheduling method and device, and relates to the technical field of communication, the method is applied to a management center, and the method comprises the steps: receiving an alarm message sent by a firewall device, the alarm message being that the occupancy rate of the firewall device in a first CPU core reaches a preset alarm threshold in a continuous preset period, the firewall device is sent under the condition that the occupancy rate of the second CPU core with the lowest occupancy rate in the firewall device is lower than a preset alarm threshold; acquiring flow monitoring information of each data flow received by the first CPU core and reported by the firewall equipment and an overload reason of the first CPU core, and identifying elephant flows in each data flow based on the flow monitoring information of each data flow; determining an adjustment strategy for the elephant flow based on the overload reason; and issuing the adjustment strategy to the firewall device, so that the firewall device schedules the elephant flow based on the adjustment strategy. According to the scheme, on the premise that the performance pressure of firewall equipment is not increased, the elephant flow is scheduled, and overload of the CPU core is avoided.

The invention provides an elephant stream level-to-level management chip, an elephant stream level-to-level management method and electronic equipment. A message transfer-in module sends a received first type of messages to an elephant stream identification module; when the elephant flow identification module receives a first type of message, the elephant flow identification module adds a classification label to the first type of message according to a target flow rate and sends the first type of message to the QoS strategy module; if the length of the messages stored in the target queue exceeds the target waterline, discarding the first type of messages, and if the length does not exceed the target waterline, storing the first type of messages in the target queue; and the QoS strategy module transmits the first type of messages in the M queues to the message transfer-out module according to a preset priority rule. As the waterline corresponding to the high-level classification label is lower, the high-level classification label is easier to discard, the occupation of the message on the cache space can be prevented, the influence on the message of the low-level classification label is reduced, and the forwarding time delay and efficiency of the message of the low-level classification label are ensured.

The invention provides a multi-core processing elephant flow order-preserving forwarding method and related equipment, and the technical scheme of the invention is as follows: after a network interface receives an elephant flow, an order-preserving mark in message description information of each message is set to be valid, and then each message description information is written into a plurality of input cache queues according to progressive increase polling of serial numbers; each CPU core of the multi-core processor obtains the message description information from the multiple input cache queues through the exclusive binding relation and performs parallel processing, and after parallel processing is completed, the message description information is forcibly written into the multiple output cache queues according to the same serial number; and finally, performing order-preserving or non-order-preserving output processing in the polling output cache queue based on the order-preserving mark. According to the elephant flow order-preserving forwarding method, the parallel processing capacity of the multi-core processor can be fully utilized, additional reordering cache resources are not added, ordered sending of elephant flow messages is ensured, and the network resource utilization rate and the data transmission efficiency are improved.

The application discloses a kind of elephant flow identification and processing method in distributed energy security data acquisition device, it is related to data stream transmission technical field, and the network traffic data is obtained by flow acquisition equipment, and the characteristic value of data packet in unit time is counted by sliding window, data stream coefficient is calculated according to data packet characteristic value to judge to generate first elephant flow processing signal, simultaneously, CPU usage rate prediction model is constructed based on convolutional neural network, and real-time CPU usage rate is substituted into CPU usage rate prediction model to output expected CPU usage rate, to judge to generate second elephant flow processing signal, and the hierarchical result is obtained by hierarchical response mechanism to elephant flow, and differential processing strategy is executed according to hierarchical result, wherein, multi-channel shunt transmission is used to primary elephant flow, dynamic encryption transmission is started to secondary elephant flow, local cache is triggered to tertiary elephant flow, and security early warning signal is generated to execute security early warning mechanism.

A networking device and system are described, among other things. An illustrative system is disclosed to include a shared buffer and at least a flow controller. In some embodiments, the system and / or flow controller may be configured to measure a packet flow's bandwidth consumption of the shared buffer, assign a flow-type attribute to the packet flow based on the packet flow's bandwidth consumption of the shared buffer, select a shared buffer schema for the packet flow based on the flow-type attribute assigned to the packet flow, and apply the selected shared buffer schema to the packet flow. For example, the flow-type attribute assigned to the packet flow may comprise a mice flow state or an elephant flow state, and a reserve attribute may be assigned to the flow based on the packet flow being assigned the mice flow state or the elephant flow state.

In some embodiments, a method receives a packet for a flow associated with a workload. Based on an indicator for the flow, the method determines whether the flow corresponds to one of an elephant flow or a mice flow. Only when the flow is determined to correspond to an elephant flow, the method enables a hardware acceleration operation on the packet. The hardware acceleration operation may include hardware operation offload, receive side scaling, and workload migration.

A semiconductor chip for implementing aggregated flow detection and management includes a number of pipes, where each pipe is coupled to a portion of ports on the semiconductor chip that are to receive data packets. A logic is coupled to the pipes and is used to detect and manage an elephant flow. The elephant flow-detection and management logic includes a flow table and a byte counter.

The invention discloses a traffic mirroring method and system based on a Bloom filter, and the method comprises the following steps: S1, network equipment copies original traffic to generate mirroring traffic; s2, judging whether to start the Bloom filter or not according to configuration requirements, and if yes, executing S3; if not, executing S4; s3, executing Bloom filter check on the mirror image traffic; s4, performing traditional sampling mirroring on the mirroring traffic passing through the Bloom filter according to a preset sampling ratio to obtain a final mirroring traffic message; and S5, outputting a final mirror image message and monitoring equipment. According to the method, the Bloom filter is introduced into the mirroring, and mirroring can be performed according to the message characteristics, so that the mouse flow can be ensured not to be ignored, and the problem of impact of the elephant flow on the network can be solved. The method provided by the invention has relatively strong practical application significance.

The invention relates to workload aware data packet booting. Systems and methods herein are used for network communications in which at least one processor may include or be associated with a datapath module and a packet steering module in which the datapath module may receive communications that may be associated with different communication protocols, the communication may receive a size indication associated with the communication and may determine an elephant stream based in part on the size indication associated with the communication, and wherein the packet steering module may receive information that may be associated with the elephant stream and may enforce at least one rule, to direct incoming packets to different receive queues associated with respective ones of the communication protocols.

This application provides a scheduling method, apparatus, electronic device, and medium, relating to the field of transmission and bearer network technology. The method includes: upon determining that a detected new connection event is an elephant flow, obtaining the source node, destination node, and bandwidth requirement of the elephant flow; pruning the path from the source node to the destination node according to the bandwidth requirement, and determining a candidate path set based on a four-factor weighted cost function of link occupancy, remaining bandwidth, physical hop count, and safety margin; calculating the incremental index of the first path in the candidate path set through sandbox simulation, and determining the first path as the target path if the incremental index passes security verification; and scheduling the elephant flow based on the target path. This reduces congestion caused by the default shortest path from the source, ensures low-latency transmission of small flows, and thus improves overall service performance.

The application discloses a flow label-based traffic scheduling method, device and equipment. The method comprises the following steps: a source computing node divides the service flow in a single RDMA queue pair into multiple independent sub-flows based on a preset semantic boundary; the source computing node generates an IPv6 flow label for each sub-flow, and fills the IPv6 flow label into the corresponding field of an IPv6 header when encapsulating a data packet; and the source computing node and / or a network node schedules the data packets of different sub-flows to different network paths for transmission based on the IPv6 flow label in a path selection process. The application can greatly improve the traffic entropy of the network by performing semantic subdivision on the service flow in the queue pair, dividing the service flow into multiple independent sub-flows, and allocating different IPv6 flow labels to each sub-flow, so that the elephant flow can be transmitted through multiple paths, and the out-of-order problem caused by the multi-path transmission can be avoided by using the intermittent characteristics of the traffic.

Systems and methods herein are for network communications, where at least one processor can include or be associated with a data path module and a packet steering module, where the data path module can receive communications that may be associated with different communication protocols and can determine an elephant flow based in part on a size indication associated with the communications, and where the packet steering module can receive information that may be associated with the elephant flow and can enforce at least one rule to steer incoming packets for different receive queues associated with respective ones of the communication protocols.

This invention discloses a network state-aware congestion-prone flow rerouting method and system. The method specifically includes: under congestion conditions, statistically calculating the contribution of each flow to the queue, dynamically determining a large flow judgment threshold based on the number of active flows, and identifying flows with contributions exceeding the large flow judgment threshold as responsible elephant flows; based on the responsible elephant flow list and network state information, modeling the rerouting problem as a mixed-integer programming model with the objective of minimizing maximum link utilization, and solving for the globally optimal path allocation scheme; according to the globally optimal path allocation scheme, installing new path forwarding rules and deleting old path rules on each switch using a build-then-tear strategy. This invention achieves sub-second congestion response and globally optimal traffic scheduling through cross-layer collaboration between the data plane and control plane, effectively avoiding the congestion migration problem caused by traditional reactive rerouting.

The invention provides a rapid elephant stream forwarding method and device, and relates to the technical field of computers, in particular to the technical field of elephant stream forwarding. A specific embodiment of the method comprises the following steps: acquiring a message of flow reaching a gateway machine; performing flow frequency statistics on the message of the flow to obtain a statistical value; in response to the statistical value exceeding a preset threshold value, determining the flow as a candidate elephant flow; identifying elephant streams from the candidate elephant streams; and forwarding the elephant flow through the fast path.

The invention provides a message forwarding method, which comprises the following steps: receiving a first flow, and redirecting the first flow to a target ECMP group for forwarding in response to the fact that the first flow is determined as an elephant flow at a flow entrance of a default ECMP group; wherein the target ECMP group is an equivalent routing group used for forwarding elephant flow; the target ECMP group comprises at least two first links, and the bandwidth of at least one first link is greater than the bandwidth of each link in the default ECMP group; by judging the elephant flow of the flow, the elephant flow is forwarded through the target ECMP group comprising the link with larger bandwidth, so that the risk of network congestion can be reduced, and the stability of the network is improved. The invention further provides electronic equipment, a computer readable medium and a computer program product.

The invention discloses a fine-grained network measurement structure based on funnel type Sketch and a measurement algorithm thereof. The fine-grained network measurement structure comprises a top layer structure, a middle layer structure and a bottom layer structure. The top layer structure is composed of a two-dimensional counter array, small counters are adopted, each row corresponds to an independent hash function, and the top layer structure is used for recording and filtering mouse flow. The middle layer structure is composed of a two-dimensional array, and each barrel comprises two counter arrays and a key value pair. Wherein the two counter arrays adopt counters with different sizes and are used for dividing the network flow more finely, and the key value pair is used for recording the flow identifier of the elephant flow and the flow size corresponding to the flow identifier of the elephant flow. The bottom layer structure is composed of a hash table and used for recording the flow overflowing from the middle layer structure. According to the method, aiming at the network flow with deflected distribution, the measurement precision of the mouse flow can be improved on the premise of ensuring the detection accuracy of the elephant flow, and rapid network flow insertion and query are realized.

Systems and methods herein for receive-buffer allocation in a network include at least one network interface controller (NIC) to handle receive requests for communication associated with a device. The at least one NIC can provide a fast path and a slow path for the communication. The slow path may be used for the communication based in part on a rule programmed in the at least one NIC, while one or more further rules in the at least one NIC can enable all of an available receive-buffer allocation for a buffer of the system based in part on burst or elephant flow in the communication, and can enable a threshold receive-buffer allocation that is less than the available receive-buffer allocation for the buffer based in part on at least one predetermined condition.

Some embodiments provide a system that detects whether a data flow is an elephant flow; and if so, the system treats it differently than a mouse flow. The system of some embodiments detects an elephant flow by examining, among other items, the operations of a machine. In detecting, the system identifies an initiation of a new data flow associated with the machine. The new data flow can be an outbound data flow or an inbound data flow. The system then determines, based on the amount of data being sent or received, if the data flow is an elephant flow. The system of some embodiments identifies the initiation of a new data flow by intercepting a socket call or request to transfer a file.