CN115632986A - Path load balancing method and system - Google Patents

Abstract

The invention relates to the technical field of data flow, in particular to a path load balancing method and a path load balancing system. The method comprises the steps of firstly calculating the residual bandwidth of each preselected path, and then selecting a target path from each preselected path according to the size of the residual bandwidth. When the newly arrived current data packet to be transmitted is different from the data packet transmitted at the last moment, the current data packet to be transmitted is transmitted on a new target path instead of transmitting the current data packet to be transmitted by using a historical path. The invention selects one path to transmit the new current data packet to be transmitted according to the residual bandwidth of each path, thereby realizing the load balance of each path. In addition, the invention does not reselect the transmission path for each data packet, but only selects a new target path for the current data packet to be transmitted, which is different from the transmitted data packet, so as to prevent the data packet received by the receiving end from being out of sequence easily caused by putting the same data packet on different paths for transmission.

Description

A method and system for path load balancing

technical field

The invention relates to the technical field of data flow, in particular to a path load balancing method and system.

Background technique

Under the condition of multi-path transmission, the situation of network asymmetry is very common (the path is the link, which is used to transmit the data flow, and the network asymmetry means that the data flow generated by the network that generates the data flow is different in different time periods), for example : Heterogeneous network, failure of intermediate network equipment or link failure, etc. In addition, due to the diversity of services on the network, different flows have different traffic characteristics, such as the arrival time of application traffic, traffic distribution, traffic size, and duration. These flows with different characteristics have different requirements on the network. How to adapt the paths allocated to data flows to a complex network environment with a variety of different traffic characteristics is an extremely important challenge. Improper load balancing (load balancing means to balance the size of data flows loaded by each path) strategy will lead to Some links are idle and the rest of the links are congested or even packet loss, or the packets of a single flow are out of order at the rendezvous point to increase the processing delay. In addition, it is necessary to have the ability to sense whether the link is invalid, so as to timely schedule the original link to the failed link Route traffic is scheduled to available links.

To sum up, the existing path load balancing is easy to cause path congestion.

Therefore, the prior art still needs to be improved and improved.

Contents of the invention

In order to solve the above technical problems, the present invention provides a path load balancing method and system, which solves the problem that the existing path load balancing easily causes path congestion.

To achieve the above object, the present invention adopts the following technical solutions:

In a first aspect, the present invention provides a path load balancing method, which includes:

Calculating the remaining bandwidth of each preselected path;

Screening out a target path from each of the pre-selected paths according to the remaining bandwidth of each of the pre-selected paths;

switch the path corresponding to the current data packet to be transmitted whose difference information satisfies the set condition from the historical path to the target path, the difference information is used to characterize the difference between the current data packet to be transmitted and the transmitted data packet , the historical path is the path of the transmitted data packet.

In an implementation manner, the screening manner of each of the preselected paths includes:

Send several probe data packets to each candidate path;

Calculating the detection duration and/or the corresponding packet loss rate used by each of the candidate paths to transfer the received detection data packets to the same target terminal;

Each of the pre-selected paths is selected from each of the candidate paths according to the detection duration and/or the packet loss rate corresponding to each of the candidate paths.

In an implementation manner, the calculating the remaining bandwidth of each pre-selected path includes:

Counting the number of received data packets and the number of sent data packets of each of the preselected paths within the measurement period;

Calculate the remaining bandwidth of each of the pre-selected paths according to the number of received data packets and the number of sent data packets of each of the pre-selected paths and the measurement period.

In an implementation manner, calculating the remaining bandwidth of each of the pre-selected paths according to the number of received data packets and the number of sent data packets of each of the pre-selected paths and the measurement period includes :

Obtaining the number of bytes received by each of the preselected paths according to the number of received data packets and the bit value of each received data packet;

Obtain the number of bytes sent for each of the preselected paths according to the number of sent data packets and the bit value of each sent data packet;

dividing the sum of the number of bytes received and the number of bytes sent by the measurement period to obtain an intermediate result;

The intermediate result is subtracted from the original bandwidth of each of the pre-selected paths to obtain the remaining bandwidth of each of the pre-selected paths.

In an implementation manner, the calculation method of the measurement period includes:

Calculate each round-trip delay of the data packet on each of the candidate paths;

The measurement period is obtained according to each of the round-trip delays and the total number of paths formed by each of the candidate paths.

In an implementation manner, the filtering out target paths from each of the pre-selected paths according to the remaining bandwidth of each of the pre-selected paths includes:

The remaining bandwidth of each of the pre-selected paths is compared, and the pre-selected path with the largest remaining bandwidth is used as the target path.

In an implementation manner, the path corresponding to the current data packet to be transmitted whose difference information satisfies the set condition is switched from the historical path to the target path, and the difference information is used to characterize the current data packet to be transmitted The difference between the transmitted data packet and the historical path is the path of the transmitted data packet, including:

calculating the flow difference between the data flow where the current data packet to be transmitted is located and the data flow where the transmitted data packet is located in the difference information;

When the flow difference satisfies the two different data flows in the set condition, switching the path of the current data to be transmitted from the historical path to the target path.

In an implementation manner, the calculating the flow difference between the data flow where the current data packet to be transmitted in the difference information is located and the data flow where the transmitted data packet is located includes:

Calculating the packet identifier of the currently to-be-transmitted data packet and the packet identifier of the transmitted data packet, where the packet identifier is used to represent a data flow label;

According to the message identifier of the current data packet to be transmitted and the message identifier of the transmitted data packet, calculate the difference between the data stream where the current data packet is to be transmitted and the data stream where the transmitted data packet is located flow difference.

In an implementation manner, the path corresponding to the current data packet to be transmitted whose difference information satisfies the set condition is switched from the historical path to the target path, and the difference information is used to characterize the current data packet to be transmitted The difference between the transmitted data packet and the historical path is the path of the transmitted data packet, including:

When the current data packet to be transmitted and the transmitted data packet belong to the same data stream, count the time between the time when the current data packet to be transmitted arrives at the packet processor and the time when the transmitted data packet arrives at the packet processor time interval;

When the time interval is greater than the time threshold in the set condition, switch the path of the currently to-be-transmitted data packet from the historical path to the target path.

In the second aspect, the embodiment of the present invention also provides a path balancing load system, wherein the system includes the following components:

a controller, configured to calculate the remaining bandwidth of each pre-selected path and filter out a target path from each of the pre-selected paths according to the remaining bandwidth of each of the pre-selected paths;

The packet processor is bidirectionally electrically connected to the controller, and is used to switch the path corresponding to the currently to-be-transmitted data packet whose difference information satisfies the set condition from the historical path to the target path.

In one implementation, the controller includes:

The topology monitoring module is used to monitor the network topology of the internal network generating the data packet;

A link detection module, configured to send several detection packets to each candidate path, the candidate path including the pre-selected path;

A link analysis module, bidirectionally electrically connected to the link detection module, configured to analyze the quality of each of the candidate paths according to the transmission status of the detection data packets on each of the candidate paths;

The path decision module is electrically connected bidirectionally with the link analysis module, and is used to obtain a preselected path according to the quality of each of the candidate paths, and obtain a target path according to the remaining bandwidth of the preselected path.

In a third aspect, an embodiment of the present invention further provides a terminal device, wherein the terminal device includes a memory, a processor, and a path load balancing program stored in the memory and operable on the processor, the When the processor executes the path load balancing program, the steps of the path load balancing method described above are implemented.

In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, where a path load balancing program is stored on the computer-readable storage medium, and when the path load balancing program is executed by a processor, the above-mentioned The steps of the path load balancing method.

Beneficial effects: the present invention first calculates the remaining bandwidth of each pre-selected path, and then selects a target path from each pre-selected path according to the size of the remaining bandwidth. When the newly arrived current data packet to be transmitted is different from the data packet transmitted at the last moment, the current data packet to be transmitted is not transmitted using the historical path, but the current data packet to be transmitted is transmitted on the new target path. It can be seen from the above analysis that the present invention selects one of the paths to transmit a new data packet currently to be transmitted according to the remaining bandwidth of each path, so as to realize load balancing of each path. And the present invention is not all to re-select the transmission path for each data packet, but only selects a new target path for the current data packet to be transmitted different from the transmitted data packet, which is to prevent the same data packet from being placed in the Transmission on different paths can easily cause out-of-sequence data packets received by the receiving end.

Description of drawings

Fig. 1 is the overall flowchart of the present invention;

FIG. 2 is a schematic diagram of network deployment in an embodiment of the present invention;

FIG. 3 is a structural diagram of a load balancing system in an embodiment of the present invention;

FIG. 4 is a flow chart of implementing load balancing in an embodiment of the present invention;

FIG. 5 is a functional block diagram of an internal structure of a terminal device provided by an embodiment of the present invention.

Detailed ways

The technical solutions in the present invention are clearly and completely described below in conjunction with the embodiments and the accompanying drawings. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

After research, it is found that under the condition of multi-path transmission, the situation of network asymmetry is very common (the path is the link, which is used to transmit the data flow, and the network asymmetry means that the data flow generated by the network that generates the data flow is different in different time periods. Same), for example: heterogeneous network, failure of intermediate network equipment or link failure, etc. In addition, due to the diversity of services on the network, different flows have different traffic characteristics, such as the arrival time of application traffic, traffic distribution, traffic size, and duration. These flows with different characteristics have different requirements on the network. How to adapt the paths allocated to data flows to a complex network environment with a variety of different traffic characteristics is an extremely important challenge. Improper load balancing (load balancing means to balance the size of data flows loaded by each path) strategy will lead to Some links are idle and the rest of the links are congested or even packet loss, or the packets of a single flow are out of order at the rendezvous point to increase the processing delay. In addition, it is necessary to have the ability to sense whether the link is invalid, so as to timely schedule the original link to the failed link Route traffic is scheduled to available links.

In order to solve the above technical problems, the present invention provides a path load balancing method and system, which solves the problem that the existing path load balancing easily causes path congestion. In the specific implementation, first calculate the remaining bandwidth of each pre-selected path, then filter out the target path according to the remaining bandwidth, and finally transmit the current data packet to be transmitted that is different from the transmitted data packet on the target path, so as to solve the problem of path congestion .

For example, as shown in FIG. 2 , the data stream (video stream or audio stream) of the internal network needs to be transmitted to the external network through a pre-selected path in the form of data packets. When one of the data packets of the data stream (the current data packet to be transmitted, denoted as data packet A) arrives at the P4 switch (packet processor), it is first judged by the P4 switch whether there is a difference between the data packet A and the data packet B (the so-called difference is Do these two data packets belong to the same data flow? If they belong to the same data flow, do these two data packets belong to the same packet set? A predetermined time threshold), if there is a difference, the data packet A is no longer transmitted over the path used to transmit the data packet B. At this time, it is necessary to use the path with the largest remaining bandwidth as the target path to transmit the data packet A according to the remaining bandwidth of each path. On the contrary, if the data packet A is transmitted on a path with a small remaining bandwidth, the congestion of the path will be aggravated. Therefore, this embodiment balances the load of each path to reduce the congestion. In addition, this embodiment does not change the transmission path of other data packets that are the same as data packet B, but keeps them on the same path as data packet B, so that the external network can receive the data packet and summarize it with data packet B together.

exemplary system

This embodiment provides a path balancing load system, as shown in Figure 3, the system includes the following components:

a controller, configured to calculate the remaining bandwidth of each pre-selected path and filter out a target path from each of the pre-selected paths according to the remaining bandwidth of each of the pre-selected paths;

The packet processor (P4 switch) is electrically connected to the controller bidirectionally, and is used to switch the path corresponding to the currently to-be-transmitted data packet whose difference information satisfies the set condition from the historical path to the target path.

In one embodiment, the controller includes:

The topology monitoring module is used to monitor the network topology of the internal network generating the data packets;

A link detection module, configured to send several detection packets to each candidate path, the candidate path including the pre-selected path;

A link analysis module, bidirectionally electrically connected to the link detection module, configured to analyze the quality of each of the candidate paths according to the transmission status of the detection data packets on each of the candidate paths;

The path decision module is electrically connected bidirectionally with the link analysis module, and is used to obtain a preselected path according to the quality of each of the candidate paths, and obtain a target path according to the remaining bandwidth of the preselected path.

In one embodiment, the P4 switch and the controller realize the load balancing of each path in FIG. 2 under the SDN architecture, and the controller is responsible for collecting the network topology information in FIG. 2 and performing link detection (detecting whether each path is suitable for transmitting data packets) and calculate the delay of each link. The P4 switch is responsible for implementing the forwarding logic formulated by the controller for the data packets on each path.

The process of the controller and the P4 switch working together to achieve load balancing of each path is as follows:

The topology monitoring module is responsible for starting and stopping the switch port or disconnecting the link, and updating the topology of the entire network (internal network) in real time. In addition, this module also maintains the multipath information between node pairs in the network topology, and transmits the multipath information to the link Road detection module and link analysis module.

The link detection module is responsible for sending a set of detection data packets to detect the quality of all paths with a fixed period (measurement period). Sending and receiving, recording the sending time stamp and the time stamp of the probe data packet received by the switch on the external network side, counting the number of data packets actually received by the switch on the sending and external network side, and passing the result to the link analysis module.

The link analysis module calculates the RTT (the round-trip time of the detection data packet on the path) through the timestamp sent and received by the link detection module, and calculates the packet loss rate by the number of packets sent and actually received, and calculates the packet loss rate according to the RTT and packet loss rate. Analyze link quality to detect faulty links and reschedule traffic in time. In the path decision-making module, the remaining bandwidth of each path is calculated by obtaining the counter data (the number of bytes sent and received by each path) of the P4 switch in a measurement period, and combined with the results of the link analysis module to perform traffic scheduling (for arriving at the P4 switch select the path with the largest remaining bandwidth).

In this embodiment, the forwarding logic based on flowlet (a collection of data packets) is realized by using a P4 switch on the data plane. The flowlet detection module first performs a hash operation based on the header of each data packet, and performs hashing on the network layer for ipv4 packets according to the five-tuple (IP source address, IP destination address, TCP source port, TCP destination port, protocol type) , for packets whose network layer is ipv6, the flow label is used for hashing. Afterwards, the time interval between data packets in the same network flow is detected according to the flowlet timeout threshold. When the detected time interval is greater than the timeout threshold (time threshold), it indicates that a new flowlet has appeared. The flowlet scheduling step is based on the remaining bandwidth of each port of the switch The principle of maximum priority schedules the next flowlet.

exemplary method

The path load balancing method in this embodiment can be applied to a terminal device, and the terminal device can be a terminal product with a computing function, such as a computer. In this embodiment, as shown in FIG. 1, the path load balancing method specifically includes the following steps S100, S200, and S300:

S100. Calculate the remaining bandwidth of each preselected path.

Step S100 includes two parts, the first part: selecting paths with good quality from each candidate path in Fig. 2 as the pre-selected path; the second part: calculating the remaining bandwidth of each pre-selected path.

The first part includes the following steps S101, S102 and S103:

S101. Send several detection data packets to each candidate path.

S102. Calculate the probing duration and/or the corresponding packet loss rate used by each of the candidate paths to forward the received several probing data packets to the same target terminal.

S103. Select each of the preselected paths from each of the candidate paths according to the detection duration and/or the packet loss rate corresponding to each of the candidate paths.

The link detection module of the controller sends several detection data packets, for example, ten detection data packets, to each candidate path in FIG. 2 via the P4 switch. In the P4 switch, a counter is added to each entry through the P4 program. The counter records the time when the detection data packet arrives at the P4 switch (transmission time). When the detection data packet reaches the switch on the external network side through the path to be selected, the external The switch sends a feedback command, and the counter records the time when the P4 switch receives the feedback command (receiving time). According to the sending time and receiving time, the detection duration of the detection data packet on the path to be selected can be calculated. The counter of the P4 switch will record that it has received ten data packets, and the counter on the switch on the external network side will also record how many data packets it has received. If the external switch only receives eight data packets, then the packet loss rate is twenty percent. If the detection duration of a candidate path is long and the packet loss rate is also large, then the candidate path is of poor quality and cannot be used for data packet transmission. Therefore, the flow table of the path to be selected with poor quality needs to be deleted from the P4 switch, so that the P4 switch only retains the flow table of the path to be selected with good quality (preselected path).

The second part includes the following steps S104 to S108:

S104. Count the number of received data packets and the number of sent data packets of each of the preselected paths within the measurement period T.

In one embodiment, the calculation process of the measurement period T is as follows:

calculating each round-trip delay of the data packet on each of the candidate paths; and obtaining the measurement period T according to each of the round-trip delays and the total number of paths formed by each of the candidate paths.

为对N开方后下取整，T的单位为秒。RTT _i is the round-trip delay of the data packet on the i-th candidate path, N is the total number of paths,

It is rounded down after the square root of N, and the unit of T is seconds.

S105. Obtain the number of received bytes RX_Bytes of each preselected path according to the number of received data packets and the bit value of each received data packet.

The number of received packets multiplied by the bit size of each packet is RX_Bytes.

S106. According to the number of sent data packets and the bit value of each sent data packet, obtain the number of sent bytes of each of the preselected paths.

The number of packets sent multiplied by the bit size of each packet is TX_Bytes.

S107. Divide the sum of the number of received bytes and the number of sent bytes by the measurement period to obtain an intermediate result.

S108. Subtract the intermediate result from the original bandwidth of each of the pre-selected paths to obtain the remaining bandwidth of each of the pre-selected paths.

Remaining bandwidth=BW-(RX_Bytes+TX_Bytes)/T, where (RX_Bytes+TX_Bytes)/T is an intermediate result.

In one embodiment, if the difference between the remaining bandwidth calculated in the measurement period and the remaining bandwidth calculated in the previous measurement period reaches 20%, it is considered that there is a state change link, and the path_rem_bw register array is updated.

S200. Select a target path from each of the pre-selected paths according to the remaining bandwidth of each of the pre-selected paths.

In this embodiment, the preselected path with the largest remaining bandwidth is used as the target path, and then the current data packet to be transmitted is transmitted from the internal network to the external network via the target path. Selecting the path with the largest remaining bandwidth to transmit the data packet can make the bandwidth of the path fully utilized, thereby achieving balance with the bandwidth of other paths.

S300, switch the path corresponding to the current data packet to be transmitted whose difference information satisfies the set condition from the historical path to the target path, the difference information is used to represent the difference between the current data packet to be transmitted and the transmitted data packet difference, the historical path is the path of the transmitted data packet.

Step S300 includes two parts: the first part judges whether the current data packet to be transmitted needs to be transmitted on the target path, and the second part transmits on the historical path if it does not need to be transmitted on the target path, or transmits on the target path.

The first part of step S300 includes the following steps:

S301. Calculate the packet identifier of the currently to-be-transmitted data packet and the packet identifier of the transmitted data packet, where the packet identifier is used to represent a data flow label.

In this embodiment, when the P4 switch receives the current data packet to be transmitted, the header identification information of the data packet is hashed to the message identification (including the network IP address of the data packet source, the network IP address to which the data packet needs to be sent) address, source port, destination port, and protocol type).

S302. According to the message identifier of the currently to-be-transmitted data packet and the message identifier of the transmitted data packet, calculate the difference between the data stream where the current to-be-transmitted data packet is located and the data stream where the transmitted data packet is located flow difference between them.

When the message identifiers of the two data packets are different, it means that the two data packets do not belong to the same data flow, that is, the data flows of the two data packets are different, so it is necessary to reselect the path for the current data packet to be transmitted for transmission , so that the load of each path is balanced.

In another embodiment, whether there is a difference between the two data packets is not only reflected in whether the two data packets belong to the same data flow but also in the time interval between the arrival of the two data packets at the P4 switch. If the time interval is large, it means that the two data packets The two data packets do not belong to the same data packet set flowlet, or even if the two data packets belong to the same data flow, but the time interval between the two data packets arriving at the P4 switch is too large, the current data packet to be transmitted needs to be reselected Target path. This embodiment comprises the steps:

S303. When the current data packet to be transmitted and the transmitted data packet belong to the same data stream, count the time when the current data packet to be transmitted arrives at the packet processor and the time when the transmitted data packet arrives at the packet processor time interval between.

S304. When the time interval is greater than the time threshold in the set condition, switch the path of the currently to-be-transmitted data packet from the historical path to the target path.

Through the above steps S301 to S304, it is judged whether the current data packet to be transmitted needs to be re-selected for transmission, and then the second part of step S300 is executed: if there is no difference between the current data packet to be transmitted and the transmitted data packet, then there is no need to re-route. When selecting a path for transmission, the path where the last packet is located continues to transmit the current packet to be transmitted, if there is a difference between the current packet to be transmitted and the transmitted packet, the target path filtered by step S200 is used to transmit the current packet to be transmitted .

The following takes Figure 4 as an example to illustrate the overall process of path load balancing in the present invention:

The present invention sets four registers on the data plane (P4 switch): flowlet_lasttime records the time when the last data packet in the flow arrives at the switch, flowlet_id is used to identify the flowlet, path_rem_bw records the remaining bandwidth corresponding to the link, next_hop records the port of the optimal path, That is, the remaining bandwidth is the largest. The detailed steps are as follows:

(1) Start the controller and P4 switch, complete the initialization work, and maintain the normal long connection between the two.

(2) Carry out hash operation to the packet header identification information of the received data packet to obtain the identification of the message. The identification information of the IPv4 message is a five-tuple, that is, the source IP address, the destination IP address, the source IP address, and the source IP address carried in the packet header. Port, destination port, and protocol type, the identification information of the IPv6 packet is the flow label.

(3) The controller assembles the detection message including the sequence number and sends the time stamp. Send detection messages in batches to the peer end regularly through packet-in, and then the peer end returns to receive through packet-out, records the received timestamp and the number of received messages, and the detection message is a ping message assembled by itself.

(4) In step (3), if the return detection packet is not received or a large number of packets are lost, or the calculated RTT obviously rises to the point that it cannot be used normally, it indicates that the corresponding link is invalid. If the link is in use, delete it immediately The flow table is selected and the flow table with the largest remaining bandwidth is selected among the available paths as the traffic outlet to re-deliver the flow table; if the link returns to normal later, it will be re-counted as a valid link.

(5) Use the counter of the P4 switch to record the number and bit size of data packets flowing through each port. In a measurement period T, the number of bytes of data packets passing through the port can be obtained as the sum of the number of sent bytes and the number of received bytes: TX_Bytes+RX_Bytes. Then the remaining bandwidth can be obtained as: BW–(TX_Bytes+RX_Bytes)/T; where BW is the bandwidth of the link, TX_Bytes is the number of bytes sent by the corresponding port, RX_Bytes is the number of bytes received by the corresponding port, and T is the measurement period.

(6) In step (5), if the difference between the remaining bandwidth calculated in the measurement period and the remaining bandwidth calculated in the previous measurement period reaches 20%, it is considered that there is a state change link, and the path_rem_bw register array is updated.

(7) Judging whether there is a matching item in the flowlet_id register array after the header identification information of the received data packet is hashed.

(8) If there is a matching item in the flowlet_id register array, the difference between the current timestamp and the timestamp recorded in the flowlet_lasttime register array is obtained through the standard_metadata of the P4 switch, and the time interval is calculated.

(9) Judging whether the time interval in step (6) is greater than the set flowlet threshold.

(10) If the time interval in step (6) is not greater than the threshold of the set flowlet, then it can be determined that the message belongs to the same flowlet, and the timestamp of receiving the message is recorded in the corresponding flowlet_lasttime register array item.

(11) Forwarding the message according to the previous message forwarding path with the flowlet.

(12) If there is no matching item in step (7), it means that a new stream appears.

(13) For step (nine), the packet interval of the same flow is greater than the flowlet threshold, increase the value of the corresponding item in the flowlet_id register array by 1, and update the flowlet_lasttime register array item to receive the timestamp of the message; for step ( 10) Adding a flowlet_id register array and a flowlet_lasttime register array belonging to the flow after performing a hash operation on the packet header identification information of the data packet.

(14): read the path_rem_bw register array, and select the link with the largest remaining bandwidth as the flow exit.

To sum up, the present invention first calculates the remaining bandwidth of each pre-selected path, and then selects a target path from each pre-selected path according to the size of the remaining bandwidth. When the newly arrived current data packet to be transmitted is different from the data packet transmitted at the last moment, the current data packet to be transmitted is not transmitted using the historical path, but the current data packet to be transmitted is transmitted on the new target path. It can be seen from the above analysis that the present invention selects one of the paths to transmit a new data packet currently to be transmitted according to the remaining bandwidth of each path, so as to realize load balancing of each path. And the present invention is not all to re-select the transmission path for each data packet, but only selects a new target path for the current data packet to be transmitted different from the transmitted data packet, which is to prevent the same data packet from being placed in the Transmission on different paths can easily cause out-of-sequence data packets received by the receiving end.

In addition, through the SDN architecture and programmable data plane technology, the present invention can timely schedule the flow to the most reasonable path according to the link status, make full use of link resources and reduce link congestion, and greatly reduce the Processing delay caused by out-of-order packets.

Based on the foregoing embodiments, the present invention further provides a terminal device, the functional block diagram of which may be shown in FIG. 5 . The terminal equipment includes a processor, a memory, a network interface, a display screen, and a temperature sensor connected through a system bus. Wherein, the processor of the terminal device is used to provide calculation and control capabilities. The memory of the terminal device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the terminal device is used to communicate with external terminals through a network connection. When the computer program is executed by the processor, a path load balancing method is implemented. The display screen of the terminal device may be a liquid crystal display screen or an electronic ink display screen, and the temperature sensor of the terminal device is pre-set inside the terminal device for detecting the operating temperature of the internal device.

Those skilled in the art can understand that the functional block diagram shown in Figure 5 is only a block diagram of a partial structure related to the solution of the present invention, and does not constitute a limitation on the terminal equipment to which the solution of the present invention is applied. The specific terminal equipment It is possible to include more or fewer components than shown in the figures, or to combine certain components, or to have a different arrangement of components.

In one embodiment, a terminal device is provided. The terminal device includes a memory, a processor, and a path load balancing program stored in the memory and operable on the processor. When the processor executes the dual-path load balancing program, the implementation is as follows Operation instructions:

Calculating the remaining bandwidth of each preselected path;

Screening out a target path from each of the pre-selected paths according to the remaining bandwidth of each of the pre-selected paths;

switch the path corresponding to the current data packet to be transmitted whose difference information satisfies the set condition from the historical path to the target path, the difference information is used to characterize the difference between the current data packet to be transmitted and the transmitted data packet , the historical path is the path of the transmitted data packet.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented through computer programs to instruct related hardware, and the computer programs can be stored in a non-volatile computer-readable memory In the medium, when the computer program is executed, it may include the processes of the embodiments of the above-mentioned methods. Wherein, any reference to memory, storage, database or other media used in the various embodiments provided by the present invention may include non-volatile and/or volatile memory. Nonvolatile memory can include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Chain Synchlink DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it can still be Modifications are made to the technical solutions described in the foregoing embodiments, or equivalent replacements are made to some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims (13)

1. A method for path load balancing, comprising:

calculating the residual bandwidth of each preselected path;

screening out a target path from each pre-selected path according to the residual bandwidth of each pre-selected path;

and switching a path corresponding to the current data packet to be transmitted, of which the difference information meets the set conditions, from a historical path to the target path, wherein the difference information is used for representing the difference between the current data packet to be transmitted and the transmitted data packet, and the historical path is the path of the transmitted data packet.

2. The path load balancing method according to claim 1, wherein the screening of each of the preselected paths comprises:

sending a plurality of detection data packets to each path to be selected;

calculating the detection duration and/or the corresponding packet loss rate used by each path to be selected for forwarding the received detection data packets to the same target terminal;

and screening each preselected path from each path to be selected according to the detection time length and/or the packet loss rate corresponding to each path to be selected.

3. The path load balancing method according to claim 2, wherein said calculating the remaining bandwidth of each preselected path comprises:

counting the number of received data packets and the number of sent data packets of each preselected path in a measuring period;

calculating the remaining bandwidth of each of the preselected paths according to the number of the received data packets and the number of the transmitted data packets of each of the preselected paths and the measurement period.

4. A path load balancing method according to claim 3, wherein said calculating the remaining bandwidth for each of said preselected paths based on the number of said received packets and the number of said transmitted packets for each of said preselected paths and said measurement period comprises:

obtaining the number of received bytes of each preselected path according to the number of the received data packets and the bit value of each received data packet;

obtaining the number of sending bytes of each preselected path according to the number of the sent data packets and the bit value of each sent data packet;

dividing the sum of the number of received bytes and the number of sent bytes by the measuring period to obtain an intermediate result;

and subtracting the intermediate result from the original bandwidth of each preselected path to obtain the residual bandwidth of each preselected path.

5. The path load balancing method according to claim 3, wherein the calculation manner of the measurement period includes:

calculating each round-trip delay of the data packet on each path to be selected;

and obtaining the measurement period according to the round trip delay and the total number of paths formed by the paths to be selected.

6. The path load balancing method according to claim 1, wherein said screening out a target path from each of said preselected paths based on a remaining bandwidth of each of said preselected paths comprises:

and comparing the residual bandwidths of the preselected paths, and taking the preselected path with the maximum residual bandwidth as the target path.

7. The path load balancing method according to claim 1, wherein the switching a path corresponding to a current packet to be transmitted whose difference information satisfies a set condition from a historical path to the target path, where the difference information is used to characterize a difference between the current packet to be transmitted and a transmitted packet, and the historical path is a path of the transmitted packet, includes:

calculating the stream difference between the data stream of the current data packet to be transmitted and the data stream of the transmitted data packet in the difference information;

and when the stream difference meets two different data streams in the set condition, switching the path of the current data to be transmitted to the target path according to the historical path.

8. The path load balancing method according to claim 7, wherein the calculating the flow difference between the data flow in which the current data packet to be transmitted is located and the data flow in which the transmitted data packet is located in the difference information includes:

calculating the message identification of the current data packet to be transmitted and the message identification of the transmitted data packet, wherein the message identification is used for representing a data flow label;

and calculating the flow difference between the data flow of the current data packet to be transmitted and the data flow of the transmitted data packet according to the message identification of the current data packet to be transmitted and the message identification of the transmitted data packet.

9. The path load balancing method according to claim 1, wherein the switching a path corresponding to a current data packet to be transmitted whose difference information satisfies a set condition from a historical path to the target path, where the difference information is used to characterize a difference between the current data packet to be transmitted and a transmitted data packet, and the historical path is a path of the transmitted data packet, includes:

when the current data packet to be transmitted and the transmitted data packet belong to the same data stream, counting the time interval between the time when the current data packet to be transmitted reaches a packet processor and the time when the transmitted data packet reaches the packet processor;

and when the time interval is greater than the time threshold value in the set condition, switching the path of the current data packet to be transmitted from the historical path to the target path.

10. A path balancing load system, comprising:

the controller is used for calculating the residual bandwidth of each preselected path and screening out a target path from each preselected path according to the residual bandwidth of each preselected path;

and the packet processor is bidirectionally and electrically connected with the controller and is used for switching the path corresponding to the current data packet to be transmitted, the difference information of which meets the set condition, from the historical path to the target path.

11. The path balancing load system of claim 10, the controller comprising:

the topology monitoring module is used for monitoring the network topology of the internal network generating the data packet;

a link detection module, configured to send a plurality of detection data packets to each candidate path, where the candidate path includes the preselected path;

the link analysis module is bidirectionally and electrically connected with the link detection module and is used for analyzing the quality of each path to be selected according to the transmission condition of the detection data packet on each path to be selected;

and the path decision module is in bidirectional electrical connection with the link analysis module and is used for obtaining a preselected path according to the quality of each path to be selected and obtaining a target path according to the residual bandwidth of the preselected path.

12. A terminal device, characterized in that the terminal device comprises a memory, a processor and a path load balancing program stored in the memory and operable on the processor, and the processor implements the steps of the path load balancing method according to any one of claims 1 to 9 when executing the path load balancing program.

13. A computer-readable storage medium, having stored thereon a path load balancing program which, when executed by a processor, implements the steps of the path load balancing method according to any one of claims 1 to 9.

Images