TWI487330B - Network system and routing method - Google Patents

Description

Network system and routing method

This case is about an electronic system and a routing method. Especially a network system and a routing method.

With the rapid advancement of information technology, various types of networks have been widely used in people's lives, such as regional networks, the Internet, and data center networks.

In general, a network includes a plurality of nodes, such as switches or routers, which typically have a forwarding table that, according to a respective forwarding table, compares the header of the packet to another packet when it receives the packet and forwards it to the other node. The forwarding table can be set manually by the network administrator or by executing a specific algorithm. Therefore, how to design a route planning method to properly set the forwarding table of nodes in the network, and avoid network congestion and improve the reliability of the network has always been an important issue in network technology.

One aspect of the invention is a routing method. The routing method uses a controller to collect the traffic and support speed of the ports of each node (such as a switch or a router) in the network, and plans a source destination pair according to the traffic and the support speed of the ports (source-destination). The transmission path of the pair).

According to an embodiment of the invention, a routing method is applied to a network system, wherein the network system includes a plurality of nodes and a controller, each of the nodes including at least one port, and two adjacent ones of the nodes Forming a plurality of links through the ports, the routing method includes: the nodes transmitting an identification information and a support speed of the ports of the nodes to the controller; the controller receiving the identification information to construct a network topology and receiving a support speed of the ports of the nodes; the controller monitoring traffic of the ports of the nodes; the controller receiving a route planning request; receiving the route planning request Afterwards, the controller calculates a link cost of the links according to the traffic of the connection between the nodes and the support speed of the ports of the nodes; the controller according to the route planning requirement and the The network topology finds a plurality of quasi-transmission paths for a source purpose; the controller sums the link costs of the links through which the quasi-transmission paths pass to calculate separately A sum of link costs of these candidate paths; and, the controller selects the plurality of candidate paths in the smallest sum of link costs of the source object as one of the packet transmission path.

According to an embodiment of the invention, the step of monitoring the traffic of the ports of the nodes comprises: the controller receiving a first accumulated traffic of the ports of the nodes at a first time point; and the controlling At a second point in time, the device receives a second accumulated flow of the ports of the nodes.

According to an embodiment of the present invention, the step of separately calculating the link cost of the links includes: dividing, by the controller, the difference between the first and second accumulated flows by the difference between the first and second time points, Obtaining an average flow rate of the ports of the nodes; the controller subtracts the average traffic of the ports of the nodes by the support speed of the ports of the nodes, respectively, to Obtaining a remaining traffic of the port of the nodes; and the controller separately dividing the remaining traffic of the port of the nodes by the support speed of the ports of the nodes to obtain a corresponding The link cost of the links of the ports of the nodes.

According to an embodiment of the invention, the controller selects the smallest sum of the link costs in the quasi-transport paths as the source destination pair, and further includes: the controller according to the links in the quasi-transport paths The smallest sum of costs is to write at least one packet forwarding rule to a forwarding table of nodes through which the sum of the link costs is the smallest in the quasi-transport paths.

According to an embodiment of the invention, the routing method further includes: the controller setting an expiration time of the at least one packet forwarding rule, so that the source destination is updated with the packet transmission path over time.

An aspect of the present invention is a network system that collects traffic and support speeds of ports of each node in a network system through a controller, and plans a source destination according to the traffic and support speed of the ports. The transmission path.

According to an embodiment of the invention, the network system includes a plurality of nodes and a controller. Each of the nodes includes at least one port, and two of the nodes form a plurality of links through the ports, and the nodes are respectively configured to output an identification information and the connection ports of the nodes. A support speed to the controller. The controller is configured to receive the identification information to construct a network topology, and receive the support speed of the connection ports of the nodes, and monitor traffic of one of the ports, and receive a route planning request. After receiving the routing planning request, the controller calculates the traffic according to the connection of the nodes and the support speed of the connection ports of the nodes respectively. a link cost of the links, and according to the route planning requirements and the network topology, find a plurality of quasi-transmission paths for a source destination pair, and add the links through which the quasi-transport paths pass The link cost is used to separately calculate a link cost sum of the quasi-transport paths, and then select the smallest sum of the link costs in the quasi-transport paths as a packet transmission path of the source destination pair.

According to an embodiment of the invention, the controller is further configured to receive a first accumulated traffic of the ports of the nodes at a first time point, and receive the connection ports of the nodes at a second time point. And a second accumulated flow rate, wherein the difference between the first and second accumulated flow rates is divided by the difference between the first and second time points, respectively, to obtain an average flow rate of the ports of the nodes.

According to an embodiment of the invention, the controller is further configured to subtract the average traffic of the ports of the nodes by the support speed of the ports of the nodes to obtain a remaining of the ports. Traffic, and the remaining traffic of the ports of the nodes are respectively divided by the support speed of the ports of the nodes to determine the chain of the links corresponding to the ports of the nodes Road cost.

According to an embodiment of the invention, the controller is further configured to write at least one packet forwarding rule into the quasi-transport path and the minimum link cost sum according to the smallest sum of the link costs in the quasi-transmission paths. The transfer table of the passed nodes.

According to an embodiment of the invention, the controller is further configured to set an expiration time of the at least one packet forwarding rule, so that the source destination is updated with the packet transmission path over time.

In summary, by applying the above embodiment, the controller can be based on the network. Calculate the link cost of each bond in the traffic between the nodes in the road and the support speed of each port, and plan a transmission path of the source destination according to the link cost of each bond, so that the source is targeted. The packet is transmitted according to the quasi-transmission path with the smallest total link cost to balance the traffic load in the network, avoiding network congestion and improving network reliability.

The spirit and scope of the present disclosure will be apparent from the following description of the preferred embodiments of the present disclosure. Modifications do not depart from the spirit and scope of the disclosure.

The terms “first”, “second”, etc. used in this document are not specifically intended to refer to the order or order, nor are they used to limit the case. They are only used to distinguish between components or operations described in the same technical terms. .

An embodiment of the present invention is a network system that collects traffic and support speeds of connections between nodes (such as switches or routers) in a network system through a controller, and according to the connections.埠 Traffic and support speed planning a source-destination pair transmission path.

FIG. 1 is a schematic diagram of a network system 10 according to an embodiment of the invention. Network system 10 includes a controller 100 and a plurality of nodes such as N1-N5. Each node N1-N5 includes at least one port, for example, node N1 includes ports P11, P12, and node N2 includes ports P21, P22, and the like. The controller 100 is connected to each of the nodes N1-N5. Adjacent ones of the nodes N1-N5 form a link with each other through a corresponding port. For example, node N1 and The node N3 forms a link L1 through the ports 11P11 and P31, and the node N1 and the node N4 form a link L2 and the like through the ports 12P12 and P41. By adjacent nodes herein is meant that two nodes can be connected to each other via a single link. In addition, the above nodes N1-N5 may be open flow switches or routers, and the above controller may be an open flow controller.

It is to be noted that the present embodiment is described by taking the connection method in FIG. 1 as an example. However, the number of nodes, the number of connection ports, and the connection relationship between the nodes are not limited thereto.

In the present embodiment, the controller 100 is, for example, a computer. The controller 100 can issue a command to the nodes N1-N5 to output an identification information and a support speed of the ports 11P11-P52 of the nodes N1-N5 to the controller 100, respectively. Then, the controller 100 can receive the identification information to construct a network topology corresponding to the network system 10 and the connection speeds of the ports 11P11-P52 of the nodes N1-N5, wherein the network topology can mean a node. The connection relationship between N1-N5. In practice, the controller 100 can cause the nodes N1-N5 to broadcast a link layer discovery protocol (LLDP) packet to the neighboring nodes N1-N5, so that the adjacent nodes N1-N5 exchange identification information. The controller 100 can then send a command to the nodes N1-N5 to transmit the identification information of the neighboring nodes N1-N5. Thus, the controller 100 can use the identification information to learn the network corresponding to the network system 10. Road topography. On the other hand, the controller 100 can also cause the nodes N1-N5 to respectively transmit the support speeds of the ports 11P11-P52 by transmitting commands to the nodes N1-N5.

In addition, the controller 100 can be used to monitor the traffic of the ports P11-P52, for example, periodically send commands to the nodes N1-N5 to make the nodes N1-N5 The flow of its connection port P11-P52 is transmitted to the controller 100.

By receiving the support speed of the connection port P11-P52 and collecting the traffic of the port P11-P52 periodically, after receiving a route planning request, the controller 100 can select the support speed based on the connection port P11-P52 and The traffic is calculated separately for the link cost of the links L1-L6. For example, the controller 100 can determine the degree of congestion of the corresponding links L1-L6 through the support speeds of the ports 11P11-P52 and the traffic thereon, and calculate the link cost accordingly.

After calculating the link cost of the links L1-L6, the controller 100 can find a plurality of source-destination pairs according to the foregoing routing planning requirements and the network topology corresponding to the network system 10. a quasi-transmission path, and summing the link costs of the links L1-L6 through which the quasi-transmission path passes, to calculate a link cost sum of the quasi-transport paths, and then selecting the smallest sum of the link costs in the quasi-transport path as the foregoing The destination of the source is a packet transmission path. The source purpose corresponds to the aforementioned routing planning requirements.

For example, if the source node is N1 and the destination node is N2, the controller 100 can respectively find two quasi-transmission paths N1→N3→N2 in the network topology corresponding to the network system 10. N1→N4→N2, and calculate the total link cost. If the total link cost of the quasi-transmission path N1→N3→N2 is smaller than the quasi-transmission path N1→N4→N2, the controller 100 selects the quasi-transmission path N1→N4. → N2 is a packet transmission path for the aforementioned source purpose.

Through the above setting, when the controller 100 plans the transmission path of the source destination, the controller 100 can select a quasi-transmission path with the smallest total link cost as the packet transmission path to balance the traffic load in the network and avoid Network congestion and improve network reliability.

The following paragraphs will further explain the monitoring of the aforementioned traffic and the calculation of the link cost.

In an embodiment of the invention, the nodes N1-N5 may respectively store statistical data, and the statistical data may include accumulated traffic (for example, cumulative transmission byte groups) of each port 11P11-P52 on each node N1-N5. The controller 100 can periodically receive the accumulated traffic of each port 11P11-P52 on each of the nodes N1-N5, and calculate the link cost of each link L1-L6 when receiving an instruction (such as a route planning request). In other words, when receiving the command, the controller 100 divides the difference between the latest received cumulative flow rate and the newly received accumulated flow rate by the time interval between receiving the accumulated flow twice to obtain the average flow rate.

Taking the connection port P11 as an example, the controller 100 can receive the first accumulated flow rate of the connection port P11 of the node N1 at the first time point (such as the 0th second), and then control after a preset time (for example, 30 seconds). The device 100 may again receive the second accumulated flow of the port 11P11 of the node N1 at the second time point (eg, at the 30th second). At this time, if the controller 100 receives the route planning request to calculate the link cost of each link L1-L6, the controller 100 may divide the difference between the first and second accumulated traffic by the difference between the first and second time points. In order to obtain the average flow rate of the connection 埠P11. Next, the controller 100 can subtract the average traffic of the port P11 by the support speed of the port P11 to obtain the remaining traffic of the port P11, and divide the remaining traffic of the port P11 by the support speed of the port P11. Corresponding to the link cost of the link L1 connecting the 埠P11.

In this way, when the support speed of the connection port P11 is much larger than the average flow rate of the port P11, the link cost of the link L1 is small, and when the support speed of the port P11 is only slightly larger than the average flow rate of the port P11, The link L1 is in a stagnation state, and the link of the link L1 is costly.

In the same manner, the controller 100 can determine the link cost of the links L1-L6. It is noted that since the controller 100 periodically receives the accumulated traffic of each port 11P11-P52 on each node N1-N5, and calculates the average traffic volume with the latest received cumulative traffic and the newly received accumulated traffic, it is a chain. The link cost of the junctions L1-L6 can vary dynamically over time.

On the other hand, in an embodiment of the present invention, the nodes N1-N5 may each include a forwarding table for storing zero to more packet forwarding rules, and the nodes N1-N5 may forward the packets according to the packet forwarding rules. When a packet enters any of the nodes N1-N5, if the node N1-N5 does not have a packet forwarding rule corresponding to the packet, the node N1-N5 may send a routing plan request to the controller 100. In some embodiments, the route planning requirement may include information of a source destination pair, wherein the source node is, for example, a node N1-N5 that transmits routing plan requirements, and the destination node is, for example, the node N1 to which the packet is to be reached. N5.

After the controller 100 receives the routing planning request, the controller 100 can calculate the link cost L1-L6 of the link according to the traffic of the connection port P11-P52 and the support speed of the port P11-P52, and according to the route planning requirement, Corresponding to the network topology of the network system 10 and the link cost of the links L1-L6, the shortest path algorithm is performed on the aforementioned source purpose pair to find the quasi-transmission path with the smallest total link cost as the source. The destination is the packet transmission path. The controller 100 performs a shortest path algorithm (for example, a Dijkstra algorithm) on the pair of source destinations, for example, by using the link cost as a weight.

Then, the controller 100 can write the at least one packet forwarding rule corresponding to the one packet transmission path separately along the one packet transmission path. A forwarding table of all nodes through which the packet transmission path passes, so that the packet of the source destination pair can be transmitted via the packet transmission path.

For example, in a case, the node N1 transmits a route planning request to the controller 100, wherein the source destination of the route planning requirement is that the source node is the node N1 and the destination node is the node N2. When the controller 100 receives the transmission route planning request of the node N1, the controller 100 can calculate the link cost of each link L1-L6, and according to the network topology corresponding to the network system 10 and the links L1-L6. The link cost is to find the quasi-transmission path with the sum of the link costs as the packet transmission path between the node N1 and the node N2, for example, N1→N4→N2. Then, the controller 100 writes the packet forwarding rules to the nodes N1 and N4 along the packet transmission path N1→N4→N2. For example, in the forwarding table of the node N1, the controller 100 may add a packet forwarding rule. The packet of the source destination pair is sent from the connection port P12; in the forwarding table of the node N4, the controller 100 may also add a packet forwarding rule, so that the packet of the source destination pair is sent from the port P42. In this way, the packets of the foregoing source destination pair can be transmitted via the minimum link cost path.

In addition, in an embodiment, the controller 100 may set the expiration time of the packet forwarding rule. After the packet forwarding rule exceeds its expiration time, the nodes N1-N5 may remove the packet forwarding rule, and thus, in the packet forwarding rule. After the removal, when the foregoing source destination packet is re-entered into the node N1-N5, the node N1-N5 may send the route planning request to the controller 100 again because the node N1-N5 does not compare the corresponding packet forwarding rule. In order for the controller 100 to re-plan the packet transmission path of the foregoing source destination, the foregoing source destination is updated with the packet transmission path over time.

Another embodiment of the present invention is a routing method. This routing method can be applied to a network system having the same or similar structure as in the foregoing FIG. 1. For convenience of description, the following operation method is described by taking the embodiment shown in FIG. 1 as an example, but is not limited to the embodiment of FIG. 1.

It is noted that in the steps of the following methods of operation, there is no particular order unless otherwise stated. In addition, the following steps may also be performed simultaneously or overlap in execution time.

FIG. 2 is a schematic diagram of a routing method 200 according to an embodiment of the invention. Routing method 200 can include steps S0-S7.

In step S0, the nodes N1-N5 can transmit the identification information and the support speed of the ports 11P11-P52 of the nodes N1-N5 to the controller 100. In step S1, the controller 100 can receive the identification information transmitted by the nodes N1-N5 to construct a network topology corresponding to the network 10, and receive the support speeds of the ports 11P11-P52 transmitted by the nodes N1-N5. For example, the controller 100 may cause the nodes N1-N5 to broadcast a link layer discovery protocol packet to their neighboring nodes N1-N5, so that the adjacent nodes N1-N5 exchange identification information, and then the controller 100 may resend The command to the nodes N1-N5 causes the identification information of the adjacent nodes N1-N5 to be transmitted back, so that the controller 100 can use the identification information to know the network topology corresponding to the network 10. Moreover, the controller 100 can also cause the nodes N1-N5 to respectively transmit the support speeds of the ports 11P11-P52 by transmitting a command to the nodes N1-N5.

In step S2, the controller 100 can monitor the traffic of the ports 11P11-P52. For example, the controller 100 may send commands to the nodes N1-N5 periodically (e.g., every predetermined period of time) to cause the nodes N1-N5 to transmit the traffic of their ports 11P11-P52 to the controller 100.

Next, in step S3, the controller 100 can determine whether the routing plan requirements transmitted by any of the nodes N1-N5 are received, and if not, the controller 100 continuously monitors the traffic of the ports P11-P52; if so, the controller 100 calculates the link cost of the links L1 - L6 based on the flow rate of the ports 11 P11-P52 and the support speed of the ports 11 P11-P52 (step S4). For example, the controller 100 can determine the degree of congestion of the corresponding links L1-L6 through the support speeds of the ports 11P11-P52 and the current traffic on them, and calculate the link cost accordingly.

Then, in step S5, the controller 100 can find a plurality of quasi-transmission paths of a source-destination pair according to the foregoing routing planning requirements and the network topology corresponding to the network system 10, and add The link cost of the links L1-L6 through which the total transmission path passes to calculate the sum of the link costs of the quasi-transport path (step S6), and then select the smallest sum of the link costs in the quasi-transport path as the source purpose. A packet transfer path is obtained (step S7). The source purpose corresponds to the aforementioned routing planning requirements.

For example, if the source node is N1 and the destination node is N2, the controller 100 can respectively find two quasi-transmission paths N1→N3→N2 in the network topology corresponding to the network system 10. N1→N4→N2, and calculate the total link cost. If the total link cost of the quasi-transmission path N1→N3→N2 is smaller than the quasi-transmission path N1→N4→N2, the controller 100 selects the quasi-transmission path N1→N4. → N2 is a packet transmission path for the aforementioned source purpose.

Through the above steps, when the controller 100 is planning a source destination transmission path, the controller 100 can select a link cost sum to be the most A small quasi-transmission path acts as a packet transmission path to balance the traffic load in the network, avoid network congestion, and improve network reliability.

In an embodiment of the present invention, in step S2, the controller 100 may periodically receive the accumulated traffic of each port 11P11-P52 on each of the nodes N1-N5. Taking the connection port P11 as an example, the controller 100 can receive the first accumulated flow rate of the connection port P11 of the node N1 at the first time point (such as the 0th second), and then control after a preset time (for example, 30 seconds). The device 100 may again receive the second accumulated flow of the port 11P11 of the node N1 at the second time point (eg, at the 30th second).

At this time, if the controller 100 receives the route planning request transmitted by the nodes N1-N5 (step S3), the controller 100 may divide the difference between the first and second accumulated traffic by the difference between the first and second time points. In order to obtain the average flow rate of the connection 埠P11. Next, the controller 100 can subtract the average traffic of the port P11 by the support speed of the port P11 to obtain the remaining traffic of the port P11, and divide the remaining traffic of the port P11 by the support speed of the port P11. The link cost corresponding to the link L1 of the port 11P11 is obtained (step S4). In other words, the controller 100 divides the difference between the latest received cumulative flow rate and the newly received accumulated flow rate by the time interval between the two received cumulative flow rates to obtain the average flow rate of the connection port P11, and then uses the average flow rate of the connection port P11. Calculate the link cost of the link L1.

By calculating the link cost by the above steps, when the support speed of the connection port P11 is much larger than the average flow rate of the port P11, the link cost of the link L1 is small, and the support speed of the port P11 is only slightly larger than the port. When the average flow rate of P11 is reached, the link L1 is in a stagnation state, and the link cost of the link L1 is high.

Similarly, controller 100 can determine the link cost of links L1-L6. It is noted that since the controller 100 periodically receives the accumulated traffic of each port 11P11-P52 on each node N1-N5, and calculates the average traffic volume with the latest received cumulative traffic and the newly received accumulated traffic, it is a chain. The link cost of the junctions L1-L6 can vary dynamically over time.

According to an embodiment of the present invention, in step S5, after the controller 100 receives the routing planning request, the controller 100 may correspond to the network topology and the links L1-L6 of the network system 10 according to the routing planning requirements. Link cost, the shortest path algorithm for the source destination pair corresponding to the route planning requirement, and the quasi-transport path with the smallest link cost sum as the destination transmission path of the source. For example, the controller 100 can use the link cost of the links L1-L6 as the weight value, and perform the shortest path algorithm (for example, the Dijkstra algorithm) on the source destination pair to find the quasi-transmission path with the smallest link cost sum. As a source of this purpose, the packet transmission path is carried out.

Next, in step S6, the controller 100 may write at least one packet forwarding rule to the forwarding table of the node through which the packet transmission path passes, according to the packet transmission path. For details of the forwarding table and the packet forwarding rule, refer to the previous embodiment, and details are not described herein again.

In addition, in an embodiment, the routing method 200 may further include a step of setting a failure time of the packet forwarding rule by using the controller 100. After the packet forwarding rule exceeds its expiration time, the node N1-N5 may forward the packet forwarding rule. After the packet forwarding rule is removed, when the packet of the source destination pair enters the node N1-N5 again, the node N1-N5 is less than the corresponding packet forwarding rule, so the node N1-N5 The routing plan request can be sent again to the controller 100 to cause the controller 100 to again The packet transmission path is planned for the foregoing source purpose, so that the foregoing source destination is updated with the packet transmission path path over time.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present case. Anyone skilled in the art can make various changes and refinements without departing from the spirit and scope of the present case. The scope defined in the patent application is subject to change.

10‧‧‧Network System

100‧‧‧ Controller

200‧‧‧Route method

N1-N5‧‧‧ node

P11-P52‧‧‧Connector

L1-L6‧‧‧ link

S0-S7‧‧‧ steps

FIG. 1 is a schematic diagram of a network system according to an embodiment of the invention; FIG. 2 is a flow chart of a routing method according to an embodiment of the invention.

200‧‧‧Route method

S0-S7‧‧‧ steps

Claims (10)

A routing method is applied to a network system, where the network system includes a plurality of nodes and a controller, each of the nodes includes at least one port, and adjacent ones of the nodes pass through the ports. Forming a plurality of links, the routing method includes: the nodes transmitting an identification information and a support speed of the connection ports of the nodes to the controller; the controller receiving the identification information to construct a network topology And receiving a support speed of the connection ports of the nodes; the controller monitors one of the ports of the nodes; the controller receives a route planning request; after receiving the route planning request, the controller Calculating a link cost of the links according to the traffic of the connection port of the nodes and the support speed of the connection ports of the nodes; the controller is configured according to the route planning requirement and the network topology Generating a plurality of quasi-transmission paths for a source purpose; the controller sums the link costs of the links through which the quasi-transmission paths pass to calculate the quasi-transmission paths separately A sum of link costs; and the controller selects the plurality of candidate paths in the smallest sum of link costs of the source object as one of the packet transmission path. The routing method of claim 1, wherein the step of monitoring the traffic of the ports of the nodes comprises: receiving, by the controller, a first accumulated traffic of the ports of the nodes at a first time point; as well as The controller receives a second accumulated flow of the ports of the nodes at a second time point. The routing method of claim 2, wherein the step of calculating the link cost of the links separately comprises: dividing, by the controller, the difference between the first and second accumulated traffic by the first and second times, respectively a difference between the points to obtain an average flow rate of the ports of the nodes; the controller subtracts the average flow of the ports of the nodes from the support speed of the ports of the nodes, respectively, to Obtaining a remaining traffic of the ports of the nodes; and the controller separately dividing the remaining traffic of the ports of the nodes by the support speed of the ports of the nodes to obtain a corresponding The links of the links of the nodes are the cost of the link. The routing method of claim 1, wherein the step of selecting, by the controller, the smallest sum of the link costs in the quasi-transport paths as the source destination pair, the packet transmission path further comprises: the controller according to the quasi-transmission The sum of the link costs in the path is the smallest, so that at least one packet forwarding rule is respectively written into one of the nodes through which the sum of the link costs in the quasi-transport path passes. The routing method of claim 4, further comprising: the controller setting an expiration time of the at least one packet forwarding rule, so that the source destination is updated with the packet transmission path over time. A network system includes: a plurality of nodes, wherein each of the nodes includes at least one port, and adjacent ones of the nodes form a plurality of links through the ports, and the nodes are respectively used for outputting And a controller for receiving the identification information to construct a network topology and receiving the connection of the nodes Supporting the speed, and monitoring one of the ports, and receiving a route planning request, wherein after receiving the route planning request, the controller according to the connection of the nodes and the traffic of the nodes The support speed of the connection port respectively calculates a link cost of the links, and finds a plurality of quasi-transmission paths of a source destination pair according to the route planning requirement and the network topology, and sums the quasi-transmissions The link cost of the links through which the path passes to calculate a link cost sum of the quasi-transport paths respectively, and then select the smallest sum of link costs in the quasi-transport paths as the minimum One of the source and destination packet transmission path. The network system of claim 6, wherein the controller is further configured to receive a first accumulated traffic of the ports of the nodes at a first time point, and receive the nodes at a second time point. a second accumulated flow rate of the connection port, and dividing the difference between the first and second accumulated flow rates by the difference between the first and second time points, respectively, to obtain an average of the connection ports of the nodes flow. The network system of claim 7, wherein the controller is further configured to subtract the nodes from the support speed of the ports of the nodes respectively. The average traffic of the port is determined to obtain a remaining traffic of the ports, and the remaining traffic of the ports of the nodes is respectively divided by the support speed of the ports of the nodes. The link cost corresponding to the links of the ports of the nodes. The network system of claim 6, wherein the controller is further configured to write at least one packet forwarding rule into the links in the quasi-transport paths according to a minimum sum of link costs in the quasi-transport paths. One of the nodes through which the sum of costs is the smallest. The network system of claim 9, wherein the controller is further configured to set an expiration time of the at least one packet forwarding rule, so that the source destination is updated with the packet transmission path over time.