JP2016171469A - Transfer system, transfer device, transfer method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allow for generation of the transfer control information, learnt from a reception packet, in a transfer device.SOLUTION: A transfer system (2) includes a transfer controller for generating and transmitting the first transfer control information including the learning processing and transfer processing that are executed under the first conditions and including the information representative of a plurality of network identifiers (Figs. 13, 10), and transfer devices for receiving the first control information from the transfer controller and storing in a storage transfer control information section, and when a received packet satisfies the first conditions with reference to the storage section, generating the second transfer control information including the transfer processing executed under the second conditions including the network identifier of the packet, according to the learning processing and storing in the storage section, and then processing the packet according to the transfer processing (Fig. 14, 20; Figs. 15, 524, 528, 532-536).SELECTED DRAWING: Figure 1

Description

  The present invention relates to a transfer system, a transfer device, a transfer method, and a program.

  OpenFlow (OpenFlow) is a protocol for SDN (Software-Defined Networking) standardized by ONF (Open Networking Foundation). In OpenFlow, a packet transfer unit having a data transfer function (D plane) and a transfer control unit having a path control function (C plane) in a conventional network device are separated as an OpenFlow switch and an OpenFlow controller. Is done. The OpenFlow protocol is used for communication between the OpenFlow controller and the OpenFlow switch. In OpenFlow, the network administrator can control the OpenFlow switch in a desired form by appropriately designing and implementing the OpenFlow controller. To that end, the OpenFlow switch has a flow table that holds the flow entries determined by the OpenFlow controller. In the flow entry, for example, information for identifying the type of received packet or a matching condition, and an action or process for various types of packets are described.

  In a known communication system, a control server in a flow control network receives a flow entry inquiry from a flow switch. In this case, when the flow entry changes the destination MAC address of the packet, the control server transmits an ARP request to the destination IP address (L3 device) of the packet. The L2 switch that has received the response to the ARP request adds the MAC address of the L3 device indicated in the response and the port number of the port that has received the response to the MAC address table. Thereby, when the flow control network and the IP network are connected by the L2 switch, even when the flow switch changes the destination MAC address of the packet, the L2 switch prevents flooding from occurring.

  Another known communication system has a first node such as a layer 2 switch, a plurality of second nodes corresponding to OpenFlow switches, and causes the first node to perform a packet transfer operation according to a packet transfer path. And a control device. The control device sends a destination learning packet having a node downstream from the first node as a transmission source from the second node downstream of the first node in the predetermined packet transfer path toward the first node. To cause the first node to perform a packet transfer operation according to the packet transfer path. As a result, the communication system can perform packet transfer according to the intended route even in an environment in which nodes typified by layer 2 switches and nodes typified by OpenFlow switches coexist.

JP 2012-39188 A Special table 2013-5377769 gazette

  With the known OpenFlow protocol, when a large number of flow entries are configured on the OpenFlow switch, the transfer time of the flow entry from the OpenFlow controller to the OpenFlow switch is long, and the OpenFlow switch is configured. Long time.

  The inventors can shorten the setting time of the transfer device by introducing a learning process for generating new transfer control information for a newly received type of packet into the transfer control information transmitted from the transfer control device. Recognized that it was possible.

  In one aspect, an object of the present invention is to enable generation of transfer control information including a condition learned from a received packet in a transfer device.

  According to one aspect of the present invention, a transfer control device that generates and transmits first transfer control information including a learning process and a transfer process that are executed under a first condition that includes information representing a plurality of network identifiers; The first transfer control information is received from the transfer control device, stored in the storage unit, and the received packet is referred to the storage unit. When the packet satisfies the first condition, the network identifier of the packet is determined according to the learning process. There is provided a transfer system including a transfer device that generates second transfer control information including a transfer process executed under the second condition including the second transfer control information, stores the second transfer control information in a storage unit, and processes the packet according to the transfer process.

  According to one aspect of the present invention, transfer control information including conditions learned from received packets in a transfer device can be generated.

FIG. 1 shows an example of a schematic configuration of a transfer system including an OpenFlow controller and a plurality of OpenFlow switches. FIG. 2 shows an example of a schematic configuration of the OpenFlow controller. FIG. 3 shows an example of a schematic configuration of the OpenFlow switch. FIG. 4 shows an example of a schematic configuration of the processor of the OpenFlow controller. FIG. 5 shows an example of a schematic configuration of the processor of the OpenFlow switch. FIG. 6 shows an example of the structure of a control packet including an OpenFlow message transmitted and received between the OpenFlow controller and the OpenFlow switch. FIG. 7A shows an example of the structure of a flow MOD message as an open flow message. FIG. 7B shows an example of the structure of the action OFPAT_LEARNED (learning action) described in the action field of FIG. 7A. FIG. 7C shows an example of the structure of the action OFPAT_EXPERIMENTER (experimental action) described in the action field of FIG. 7A. (Explained in Fig. 7A) (Explained in Fig. 7A) FIG. 8 shows an example of the structure of a FLOW_LEARNED (learning content notification) message as an OpenFlow message. 9A to 9D show an example of packet processing by a flow table including a plurality of flow entries in an ordinary form in the OpenFlow switch. (Explained in Figure 9A) FIGS. 10A to 10C show an example of packet processing by another flow table including one flow entry representing a plurality of addresses in an ordinary form in the OpenFlow switch. FIGS. 11A to 11C show an example of packet processing by a flow table including a learning type flow entry in the embodiment in the OpenFlow switch. (Explained in Figure 11A) (Explained in Figure 11A) 12A to 12D show an example of packet processing by the flow table including another learning type flow entry in the embodiment in the OpenFlow switch. (Explained in Fig. 12A) (Explained in Fig. 12A) (Explained in Fig. 12A) FIG. 13 shows an example of a flowchart of a process for generating and transmitting a flow entry including a learning algorithm, which is executed by the OpenFlow controller. FIG. 14 shows an example of a flowchart of a process executed by the OpenFlow switch for receiving a flow entry including a learning algorithm and setting the flow entry in the flow table. FIG. 15 shows an example of a flowchart for processing a received packet based on a flow table executed by the OpenFlow switch. FIG. 16 shows an example of a flowchart of processing for updating the flow table for the OpenFlow switch in the OpenFlow controller, which is executed by the OpenFlow controller. FIG. 17 shows an example of a flowchart of processing for collecting statistical information from the OpenFlow switch, which is executed by the OpenFlow controller.

The objects and advantages of the invention will be realized and attained by means of the elements and combinations particularly pointed out in the appended claims.
It is understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not intended to limit the invention.

  Non-limiting embodiments of the present invention will be described with reference to the drawings. In the drawings, similar components and elements have the same reference numerals.

  SDN (Software-Defined Networking) has a structure in which an OpenFlow controller in a control layer or C plane controls an OpenFlow switch in an infrastructure layer or D plane. Communication between the OpenFlow controller and each OpenFlow switch is performed via the control line network using, for example, the OpenFlow protocol as a control protocol. On the other hand, in order to cope with the detailed bandwidth control and security enhancement needs of the Internet, it may be required to collect statistical information for each user and for each server providing a service. In OpenFlow, the network manager creates a flow entry for each source IP address and stores it in the flow table in order to obtain statistical information of packets passing through the OpenFlow switch for each source IP address. Set in advance.

  On the other hand, when the network administrator sets, for example, the flow entry for each user in the OpenFlow switch and collects the statistical information of each user, the number of flow entries to be set increases. As the number of flow entries set in the OpenFlow switch increases, the time required for setting the OpenFlow switch increases. Then, if the OpenFlow switch fails while the OpenFlow switch is in operation, the time during which the packet cannot be transferred or the interruption time of the transfer service becomes longer. Also, if the number of flow entries to be set in the OpenFlow switch increases, the bandwidth of the control line required when setting the flow entry increases, so it is necessary to secure a wide bandwidth of the control line in advance. The maintenance cost of the control line is high.

  The inventors have expanded the OpenFlow protocol to create a flow entry including a learning process for creating a new flow entry obtained by learning a match condition from each received packet type. Recognized that it should be introduced. In this case, the learning process learns from the received packet how to transfer the received packet of a type not specified in the flow table, without inquiring the OpenFlow controller, and includes a flow entry including a matching condition obtained by learning. Is set in the flow table.

  An object of the embodiment is to enable a plurality of flow entries to be set by a learning process based on one flow entry.

  Another object of the embodiment is to shorten the setting time of the OpenFlow switch necessary for setting a plurality of flow entries.

  Yet another object of the embodiment is to further shorten the interruption time of the packet transfer service that may occur when the OpenFlow switch fails during operation of the OpenFlow switch.

  Another object of the embodiment is to reduce the control line maintenance cost by reducing the bandwidth of the control line to be secured.

  These objects can be achieved by the embodiments.

  FIG. 1 illustrates an OpenFlow controller 10 and a plurality of OpenFlow switches 20, 22,. . . An example of a schematic configuration of the transfer system 2 including 24 is shown.

  In FIG. 1, the OpenFlow controller 10 is connected to OpenFlow switches 20 to 24 through a control line network 5. The OpenFlow controller (controller) 10 is a transfer control device, and transmits a flow entry to each OpenFlow switch 20 to 24 via an L2 (Layer 2) switch 52 in the control circuit network 5. Each of the OpenFlow switches 20 to 24 is a packet transfer device, and sets the received OpenFlow entry in its own flow table.

  Each of the OpenFlow switches 20 to 24 has a plurality of communication ports connected to the network 7, and executes a corresponding action (process) according to a match condition of each flow entry in its own flow table to perform statistics. Generate information. Each of the OpenFlow switches 20 to 24 transmits a received packet to a network via a designated output port according to a match condition in the flow table according to, for example, the source and / or destination IP address of the packet received at the input port. Forward to 7. At that time, each OpenFlow switch 20 to 24 stores, for each match condition, for example, statistical information such as the number of packets or the number of bytes processed or transferred in the flow table. The statistical information may be used by a carrier or a network user of the network 7.

  The network 7 may include, for example, the Internet, ISDN, packet network, dedicated network, and the like.

  FIG. 2 shows an example of a schematic configuration of the OpenFlow controller 10.

  In FIG. 2, the OpenFlow controller 10 includes, for example, a processor 102, a storage unit 104, a communication unit 112, and an input / output unit (I / O) 116 connected via an internal bus.

  The processor 102 may be a CPU (Central Processing Unit) for a computer. The storage unit 104 includes a main storage device and an auxiliary storage device. The main storage device includes a storage device such as a semiconductor memory. The auxiliary storage device may include, for example, a semiconductor memory such as a flash memory and / or a hard disk drive (HDD).

  The input / output unit 116 can be connected to, for example, a display device such as a liquid crystal display device, an input device such as a keyboard and a pointing device, or the like via each input / output terminal 128.

  The communication unit 112 is provided with a plurality of control ports 122. The communication unit 112 is connected to the OpenFlow switches 20 to 24 through the L2 switch 52 in the control line network 5 through each control port 122.

  The OpenFlow controller 10 may be connectable to the external drive 160. The drive 160 may be for reading a recording medium 164 such as an optical disk or a magnetic disk in which software or a program is recorded. The software may include, for example, an OS, a database management system (DBMS), an application program for communication control, and the like.

  The processor 102 may be a dedicated processor implemented as an integrated circuit, for example. Further, the processor 102 may operate in accordance with an OS and a communication application program stored in the storage unit 104. The application program may be stored in the recording medium 164, read from the recording medium 164 by the drive 160, and installed in the OpenFlow controller 10.

  FIG. 3 shows an example of a schematic configuration of the OpenFlow switches 20 to 24.

  In FIG. 3, each of the OpenFlow switches 20 to 24 includes, for example, a processor 202, a storage unit 204, a communication unit 212, and an input / output unit (I / O) 216 connected via an internal bus. .

  The processor 202 may be a CPU (Central Processing Unit) for a computer. The storage unit 204 includes a main storage device and an auxiliary storage device. The main storage device includes a storage device such as a semiconductor memory. The auxiliary storage device may include a semiconductor memory such as a flash memory, for example. The processor 202 may be a dedicated processor implemented as an integrated circuit, for example.

  The input / output unit 216 can be connected to, for example, a display device such as a liquid crystal display device, an input device such as a keyboard and a pointing device, or the like via each input / output terminal 228.

  The communication unit 212 is provided with a control port 222 and a plurality of communication ports 224. The communication port 224 includes a plurality of input ports or input units and a plurality of output ports or output units, or a plurality of input / output ports or input / output units. The communication unit 212 is connected to the OpenFlow controller 10 via the control port 222 and the L2 switch 52 in the control line network 5. Further, the communication unit 212 can be connected to a node device, router device, and / or other network communication device on the network 7 via the communication port 224.

  Open flow switches 20-24 may be connectable to external drive 260. The drive 260 may be for reading a recording medium 264 such as an optical disk or a magnetic disk in which software or a program is recorded. The software may include, for example, an OS, a database management system (DBMS), a communication application program, and the like.

  The processor 202 may be a dedicated processor implemented as an integrated circuit, for example. Further, the processor 202 may operate in accordance with an OS and a communication application program stored in the storage unit 204. The application program may be stored in the recording medium 264, read from the recording medium 264 by the drive 260, and installed in the OpenFlow switches 20 to 24. As an alternative, the application program may be downloaded from the OpenFlow controller 10 or a communication server and installed in the OpenFlow switches 20 to 24, for example.

  As will be described later, the storage unit 204 of the OpenFlow switches 20 to 24 holds a flow table. Further, as will be described later, the OpenFlow controller 10 stores the flow entry transmitted to the OpenFlow switches 20 to 24 and the flow entry received from the OpenFlow switches 20 to 24 in the storage unit 104. Save to the corresponding flow table.

  FIG. 4 shows an example of a schematic configuration of the processor 102 of the OpenFlow controller 10.

  The processor 102 includes, for example, a control unit 1020, a flow entry generation unit 1024, a statistical information collection unit 1026, a flow table synchronization processing unit 1028, a learning algorithm generation unit 1030, and other processing units 1032. The control unit 1020 supplies control signals to the flow entry generation unit 1024, statistical information collection unit 1026, flow table synchronization processing unit 1028, learning algorithm generation unit 1030, and processing unit 1032 to control the operation of these elements. May be.

  The flow entry generation unit 1024 generates a flow entry to be transmitted to the open flow switches 20 to 24 according to the operation of the administrator. The statistical information collection unit 1026 collects the statistical information transmitted from the open flow switches 20 to 24 for each flow entry and stores it in the flow table for each open flow switch 20 to 24 in the storage unit 104. The flow table synchronization processing unit 1028 performs synchronization processing with the OpenFlow switches 20 to 24 so that the flow entries in the flow tables match each other. The learning algorithm generation unit 1030 generates a learning algorithm for learning processing included in the processing contents of the flow entry transmitted to the open flow switches 20 to 24.

  FIG. 5 shows an example of a schematic configuration of the processor 202 of the OpenFlow switches 20 to 24.

  The processor 202 includes, for example, a control unit 2020, a flow table reference unit 2024, a packet transfer unit 2026, a packet-in message generation unit 2028, and a flow entry setting unit 2030. The processor 202 further includes, for example, a received packet self-expanding unit 2032, a learning content notification message generating unit 2034, and other processing units 2036. The control unit 2020 may supply control signals to the flow table reference unit 2024, the packet transfer unit 2026, the packet-in message generation unit 2028, and the flow entry setting unit 2030 to control the operation of these elements. . The control unit 2020 may further supply control signals to the received packet self-expansion unit 2032, the learning content notification message generation unit 2034, and the processing unit 2036 to control the operation of these elements.

  In response to the reception of the packet, the flow table reference unit 2024 refers to the flow table in the storage unit 204 for the received packet. The packet transfer unit 2026 outputs the received packet to a designated output port in the communication port 224 and transfers it to another node on the network 7. The packet-in message generation unit 2028 generates a packet-in message for inquiring the OpenFlow controller 10 about the received packet. Here, the packet-in message is, for example, a message for inquiring the OpenFlow controller 10 about the processing method of the received packet whose processing method is not shown in the flow table. The flow entry setting unit 2030 registers or sets the flow entry received from the OpenFlow controller 10 or generated by the learning process in the flow table. Received packet self-expanding section 2032 self-expands the received packet according to the learning algorithm of the learning process, and generates a matching condition corresponding to the received packet. The learning content notification message generation unit 2034 generates a message for notifying the flow entry generated by learning and set in the flow table in order to synchronize the flow table with the OpenFlow controller 10.

  FIG. 6 shows an example of the structure of a control packet including an OpenFlow message transmitted / received between the OpenFlow controller 10 and the OpenFlow switches 20 to 24.

  In FIG. 6, the control packet includes, for example, an Ethernet (registered trademark) header, an IP header, a TCP header, and an OpenFlow message as TCP data.

  FIG. 7A shows an example of the structure of a flow MOD message as an open flow message. The flow MOD message is a flow table update request.

  The flow MOD message includes, for example, a common header part and flow MOD information. The flow MOD information includes, for example, match information and instructions information array. The instruction information array includes n instructions (information). The match information includes, for example, a match type and an oxm field. Each instruction includes, for example, an instruction type and an action field. As an example of the description of the action field according to the embodiment, there are two types of actions: OFPAT_LEARNED (learning action) and OFPAT_EXPEREMENTER (experimental action). The action OFPAT_LEARNED is a newly added action as illustrated in FIG. 7B described below. Further, the action OFPAT_EXPERIMENTER is an action proposed as an experimental additional action illustrated in FIG. 7C described below.

  FIG. 7B shows an example of the structure of the action OFPAT_LEARNED (learning action) described in the action field of FIG. 7A.

  The action OFPAT_LEARNED includes, for example, a type and a learning action (learning_action). The learning action (learning_action) includes, for example, learning_id (value of identification information indicating the learning action), priority (priority of flow), and n pieces of learning match type (learning_match_type) information. Each learning match type includes, for example, an oxm field (a value indicating the type of matching condition to be learned).

  FIG. 7C shows an example of the structure of the action OFPAT_EXPERIMENTER (experimental action) described in the action field of FIG. 7A.

  The action OFPAT_EXPERIMENTER includes, for example, an experimenter (a type number of a learning action to be applied to the ONF) in addition to each field of the action OFPAT_LEARNED in FIG.

  As described above, the learning action indicating the contents of the learning process to be executed by the OpenFlow switches 20 to 24 can be described in one of two types of actions OFPAT_LEARNED and OFPAT_Experimenter in the flow MOD message.

  FIG. 8 shows an example of the structure of a FLOW_LEARNED (learning content notification or learned flow notification) message as an open flow message. The flow LEARNED message is a message for notifying the contents of the learned flow in the open flow switches 20 to 24.

  The FLOW_LEARNED message includes, for example, a common header part and flow LEARNED information. The FLOW_LEARNED information includes, for example, learning_id (a value of identification information indicating a learning action), an OXM TLV header, and an OXM payload. The learning_id, the OXM TLV header, and the OXM payload are repeated by the length of the learning match type information array of the learning action used for learning the matching condition by the received packet.

  Next, the operation of the OpenFlow switch 20 according to the flow entry including the normal match condition and action will be described.

  9A to 9D show an example of packet processing by the flow table including a plurality of flow entries in the normal form in the OpenFlow switch 20.

  FIG. 9A shows an example of a flow table set in the OpenFlow switch 20 by the OpenFlow controller 10.

  In the flow entry of the flow table of FIG. 9A, a series of 255 source IP addresses src_ip = 192.168.1.1 / 32 to 192.168.1.255/32 is set as a match condition. To do. For this purpose, the OpenFlow controller 10 transmits 255 flow entries to the OpenFlow switch 20, and the OpenFlow switch 20 sets 255 flow entries in the flow table of FIG. 9A. Further, in each flow entry of the flow table of FIG. 9A, the output process “output = 2” to the output port of the port number “2” is performed as a corresponding action executed when the match condition is satisfied for the received packet. Is set. Here, the match condition is an execution condition of the corresponding action. The action indicates that the received packet is output to the output port having the port number “2”. In each flow entry, the initial value of the statistical information is, for example, the number of packets = 0 packets.

  In the flow table of FIG. 9A, a large number of flow entries are transmitted from the OpenFlow controller 10 to the OpenFlow switch 20 via the control line network 5. Therefore, it takes a long time to transmit the flow entry, and the bandwidth of the control circuit network 5 used for transmitting the flow entry is large. In addition, since there are many flow entries set in the flow table of the OpenFlow switch 20, the setting time of the flow entry in the OpenFlow switch 20 is also long.

  In FIG. 9B, the OpenFlow switch 20 (or its processor 202) analyzes the packet received at the input port, and the received packet satisfies any match condition in the flow entry of each row in the flow table. Determine whether or not. In this case, in the first received packet, the source IP address is src_ip = 192.168.1.1, the destination IP address is dst_ip = 192.168.10.2, and the protocol type is ip_type = 6. To do. Therefore, since the received packet satisfies the match condition src_ip = 192.168.1.1 / 32 of the flow entry in the first row of the flow table, the received packet is output to the output port with the port number “2”. Further, as statistical information of the flow entry in the first line, the number of packets = 1 packet is recorded (broken line).

  In FIG. 9C, the OpenFlow switch 20 (processor 202) analyzes the next received packet to determine whether the received packet satisfies any of the match conditions in the flow entry of each row in the flow table. judge. In this case as well, in the received packet, the source IP address is src_ip = 192.168.1.1, the destination IP address is dst_ip = 192.168.10.2, and the protocol type is ip_type = 6. . Accordingly, as in the case of FIG. 9B, the received packet is output to the output port of the port number “2”, and the number of packets = 2 packets (packet) as the statistical information of the flow entry in the first row in the flow table of FIG. 9C. Is recorded (dashed line).

  In FIG. 9D, the OpenFlow switch 20 (processor 202) analyzes the next received packet to determine whether the received packet satisfies any of the match conditions in the flow entry of each row in the flow table. judge. In this case, in the received packet, the source IP address is src_ip = 192.168.1.2, the destination IP address is dst_ip = 192.168.10.2, and the protocol type is ip_type = 6. Therefore, since the received packet satisfies the match condition src_ip = 192.168.1.2 / 32 of the flow entry in the second row in the flow table of FIG. 9C, the received packet is sent to the output port with the port number “2”. Is output. Further, as statistical information of the flow entry in the second row in the flow table, the number of packets = 1 packet is recorded (broken line).

  FIGS. 10A to 10C show an example of packet processing by another flow table including one flow entry representing a plurality of addresses in the normal form in the OpenFlow switch 20.

  FIG. 10A shows an example of a flow table set in the OpenFlow switch 20 by the OpenFlow controller 10.

  In the flow entry of the flow table of FIG. 10A, it is assumed that a range-specified source IP address src_ip = 192.168.1.0 / 24 is set as a match condition. Here, “/ 24” represents a subnet mask. Src_ip = 192.168.1.0 / 24 is a series of 255 IP addresses in the range of the source IP address src_ip = 192.168.1.1 / 32 to 192.168.1.255/32. Is included. As a result, consecutive numerical IP addresses can be simply expressed by a subnet mask. In this case, the OpenFlow controller 10 may send one flow entry to the OpenFlow switch 20, and the OpenFlow switch 20 may set one flow entry in the flow table of FIG. 10A. Also, in the flow entry of the flow table of FIG. 10A, the output process “output = 2” to the output port of the port number “2” is set as the corresponding action executed when the match condition is satisfied for the received packet. Has been. In this case, the received packet is output to the output port with the port number “2”. In the flow entry, the initial value of the statistical information is, for example, the number of packets = 0 packets.

  In FIG. 10B, the OpenFlow switch 20 (processor 202) analyzes the packet received at the input port, and whether the received packet satisfies any of the match conditions in the flow entry of each row in the flow table. Determine. In this case, in the received packet, as in FIG. 9A, the source IP address is src_ip = 192.168.1.1, the destination IP address is dst_ip = 192.168.10.2, and the protocol type is ip_type = 6. Suppose that Therefore, since the received packet satisfies the match condition src_ip = 192.168.1.0 / 24 of the flow entry in the flow table, the received packet is output to the output port with the port number “2”. As the statistical information of the flow entry in FIG. 10B, the number of packets = 1 packet is recorded (broken line).

  In FIG. 10C, the OpenFlow switch 20 (processor 202) analyzes the next received packet to determine whether the received packet satisfies any of the match conditions in the flow entry of each row in the flow table. judge. In this case, in the received packet, the source IP address is src_ip = 192.168.1.2, the destination IP address is dst_ip = 192.168.10.2, and the protocol type is ip_type = 6. Accordingly, as in the case of FIG. 9D, the received packet satisfies the match condition src_ip = 192.168.1.0 / 24 of the flow entry of the flow table, so that the received packet is output to the output port of the port number “2”. Is done. In addition, as the flow entry statistical information of FIG. 10C, the number of packets = 2 packets is recorded (broken line). Accordingly, when an address range is designated as a match condition, one piece of statistical information is generated for the entire range of address designated, and no statistical information is generated for each address.

  Next, the operation of the OpenFlow switch 20 according to the flow including the match condition and the action according to the embodiment will be described. In this case, the action includes a learning action or a learning process.

  FIGS. 11A to 11C show an example of packet processing by the flow table including the learning type flow entry in the embodiment in the OpenFlow switch 20. In this case, the learning type flow entry includes a learning action or a learning process.

  FIG. 11A shows an example of a flow table including learning type flow entries set in the OpenFlow switch 20 by the OpenFlow controller 10.

  In order to set a plurality of flow entries in the flow table of the OpenFlow switch 20, the OpenFlow controller 10 sends one learning-type flow entry or transfer control information including a learning algorithm to the OpenFlow switch 20. Send to. Next, the OpenFlow switch 20 sets one received learning type flow entry in the flow table. As a result, a flow entry is set in the flow table as shown in FIG. 11A.

  In the flow entry of the flow table of FIG. 11A, a source IP address src_ip = xx / 24 as a network identifier is set as a match condition. In the flow entry of the flow table of FIG. 11A, learning processing “learning” and output processing “output = 2” are set as corresponding actions to be executed when the match condition is satisfied.

The content of the learning process “learning” includes, for example, the following three items.
(1) “learning_id = 1”,
(2) “priority = 3000”,
(3) “learning_match_type = [OFPXMT_OFB_IPV4_SRC, OFPXMT_OFB_IPV4_DST, OFPXMT_OFB_IP_PROTO]”.

Here, “learn_id = 1” is a value of identification information indicating a learning action, and the value in this case is “1”. “Priority = 3000” is the priority of the flow generated by the learning process. The learning match type “learning_match_type” represents that a match condition for executing an action such as a learning action is described. “OFPXMT_OFB_IPV4_SRC” represents that the source address of the received packet is copied and used as part of the match condition. “OFPXMT_OFB_IPV4_DST” represents that the destination address of the received packet is copied and used as part of the match condition. “OFPXMT_OFB_IP_PROTO” represents that the protocol type of the received packet is copied and used as part of the match condition.

  In the flow table of FIG. 11A, one learning type flow entry is transmitted from the OpenFlow controller 10 to the OpenFlow switch 20 via the control line network 5. Accordingly, the time required for transmission of the learning type flow entry is short, and the bandwidth of the control circuit network 5 used for transmission of the learning type flow entry is small. Further, the number of learning type flow entries set in the flow table of the OpenFlow switch 20 is small. Therefore, the learning flow entry setting time in the OpenFlow switch 20 is also short.

  In FIG. 11B, the OpenFlow switch 20 (processor 202) analyzes the first received packet, and in order of priority of the flow entry, the received packet is one of the flow entries in each row in the flow table. Determine if match condition is met. In this case, in the first received packet, the source IP address is src_ip = xx, the destination IP address dst_ip = yy, and the protocol type is ip_type = 6. Accordingly, since the received packet satisfies the match condition src_ip = xx / 24 of the flow entry with priority 1 in the flow table, the learning action or the learning process is executed.

  In the learning action, in the flow table of FIG. 11B, a new flow entry is generated that includes a new match condition based on the source address, destination address, and protocol type of the received packet. At this time, the source address, the destination address, and the protocol type of the received packet are extracted according to the learning match type of the learning action, and written in the flow table as a new flow entry match condition.

  In this case, the match conditions for the new flow entry are the source IP address src_ip = xx / 32 as the network identifier, the destination IP address dst_ip = yy / 32 as the network identifier, and the protocol type ip_type = 6. In addition, the priority “3000” specified by the learning action and the remaining action “output = 2” of the learning type flow entry using the match condition as the operation condition are set in the new flow entry. .

  Thereby, in the OpenFlow switch 20, a new flow entry including a matching condition of each type of received packet learned from the received packet based on the contents of the learning process is generated, and the new flow entry is stored in the flow table. Can be set.

  Next, the OpenFlow switch 20 (processor 202) executes the remaining action “output = 2” other than the learning action. Therefore, the received packet is output to the output port with the port number “2” and transferred. The OpenFlow switch 20 (processor 202) further records the number of processed packets = 1 packet (broken line) as statistical information in the flow entries of the priority “1” and “3000” (broken line).

  Next, the OpenFlow switch 20 (processor 202) sends the learning content notification message including the learning action identification information, the priority, and the learned matching condition (dashed line) in the added new flow entry to the OpenFlow controller 10. Send to. The OpenFlow controller 10 (processor 102) extracts learning action identification information, priority, and learned match condition (broken line) from the received learning content notification message. Next, the OpenFlow controller 10 (processor 102) sends a flow entry including the extracted learned action identification information, priority, and learned match condition (broken line) to the flow for the OpenFlow switch 20 in the storage unit 104. Save to table.

  As a result, the flow table of the OpenFlow switch 20 and the flow table for the OpenFlow switch 20 in the OpenFlow controller 10 are matched and synchronized. Transmission of the learning content notification message of FIG. 11B for such flow table synchronization occurs when different types of packets satisfying the matching condition of the learning type flow entry are received. It occurs in a distributed manner. Therefore, the bandwidth of the control circuit network 5 used for transmitting the learning content notification message can be small. Further, even if the transmission of the learning content notification message from the OpenFlow switch 20 to the OpenFlow controller 10 is delayed, the packet transfer service provided by the OpenFlow switch 20 is not affected.

  In FIG. 11C, the OpenFlow switch 20 (processor 202) analyzes the next received packet, and the received packet matches one of the flow entries in the flow table in the order of priority of the flow entry. Determine whether the condition is met. Also in this case, in the received packet, the source IP address is src_ip = xx, the destination IP address dst_ip = yy, and the protocol type is ip_type = 6.

  Therefore, the OpenFlow switch 20 (processor 202) first compares the header and the like of the received packet with the match condition of the flow entry with the high priority “3000” generated in FIG. 11B. Since the received packet satisfies the match condition of the flow entry with the priority “3000”, the received packet is output to the output port with the port number “2” and transferred according to the corresponding action. In this case, the number of packets processed up to that point = 2 packets (packet) is recorded as statistical information in the flow entry with the priority “3000” (broken line).

  In this way, each time a new packet that satisfies the match condition src_ip = xx / 24 is received according to the learning type flow entry, the learning action is executed and a new flow entry with a high priority is generated. . Thereby, statistical information can be obtained for each flow entry added to the flow table. The statistical information for each flow entry in the flow table is later collected by the OpenFlow controller 10.

  12A to 12D show an example of packet processing by the flow table including another learning type flow entry in the embodiment in the OpenFlow switch 20.

  FIG. 12A shows an example of a flow table including another learning type flow entry set in the OpenFlow switch 20 by the OpenFlow controller 10.

  In order to set a plurality of flow entries in the flow table of the OpenFlow switch 20, the OpenFlow controller 10 sends another learning type flow entry including the learning algorithm or transfer control information to the OpenFlow switch. Transmit to the switch 20. Next, the OpenFlow switch 20 sets one received learning type flow entry in the flow table. As a result, a flow entry is set in the flow table as shown in FIG. 12A.

  Match conditions are set in the flow entry of the flow table of FIG. 12A. In this case, the match conditions are the range-specified source IP address src_ip = 192.168.1.0 / 24, the range-specified destination IP address src_ip = 192.168.10.0 / 24, and the protocol type “Ip_type = 6” is included. In this case, src_ip = 192.168.1.0 / 24 is a series of 255 IPs in the range of source IP address src_ip = 192.168.1.1 / 32 to 192.168.1.255/32. Contains an address. Also, src_ip = 192.168.10.0 / 24 is the destination IP address src_ip = 192.168.10.1 / 32 to src_ip = 192.168.10.255 / 32 Contains an address.

  Also, in the flow entry of the flow table of FIG. 12A, learning processing “learning” and output processing “output = 2” are set as corresponding actions to be executed when the match condition is satisfied. The content of the learning process “learning” is the same as in the case of FIG. 11A, for example.

  In FIG. 12B, the OpenFlow switch 20 (processor 202) analyzes the first received packet, and in order of priority of the flow entry, the received packet is one of the flow entries in each row in the flow table. Determine if match condition is met. In this case, in the first received packet, the source IP address is src_ip = 192.168.1.1, the destination IP address is dst_ip = 192.168.10.2, and the protocol type is ip_type = 6. . Therefore, since the received packet satisfies the match condition of the flow entry with the priority “1” in the flow table, the learning action is executed.

  In the learning action, a new flow entry is generated in the flow table of FIG. 12B that includes a new match condition based on the source address, destination address, and protocol type of the received packet. At this time, the source address, the destination address, and the protocol type of the received packet are extracted according to the learning match type of the learning action, and written in the flow table as a new flow entry match condition.

  In this case, the match conditions of the new flow entry are: match condition source IP address src_ip = 192.168.1.1 / 32, destination IP address src_ip = 192.168.10.2 / 32, and protocol type “ ip_type = 6 ″. In addition, the priority “3000” specified by the learning action and the remaining action “output = 2” of the learning type flow entry using the match condition as the operation condition are set in the new flow entry. .

  Thereby, in the OpenFlow switch 20, a new flow entry including a matching condition for each type of received packet learned from the received packet based on the contents of the learning process can be generated and set in the flow table.

  Next, the OpenFlow switch 20 (processor 202) executes the remaining action “output = 2” other than the learning action. Therefore, the received packet is output to the output port with the port number “2” and transferred. The OpenFlow switch 20 (processor 202) further records the number of processed packets = 1 packet (broken line) as statistical information in the flow entries of the priority “1” and “3000” (broken line).

  Next, the OpenFlow switch 20 (processor 202) sends the learning content notification message including the learning action identification information, the priority, and the learned matching condition (dashed line) in the added new flow entry to the OpenFlow controller 10. Send to. The OpenFlow controller 10 (processor 102) extracts learning action identification information, priority, and learned match condition (broken line) from the received learning content notification message. Next, the OpenFlow controller 10 (processor 102) sends a flow entry including the extracted learned action identification information, priority, and learned match condition (broken line) to the flow for the OpenFlow switch 20 in the storage unit 104. Save to table.

  As a result, the flow table of the OpenFlow switch 20 and the flow table for the OpenFlow switch 20 in the OpenFlow controller 10 are matched and synchronized.

  In FIG. 12C, the OpenFlow switch 20 (processor 202) analyzes the next received packet, and the received packet matches one of the flow entries in the flow table in the order of priority of the flow entry. Determine whether the condition is met. In this case as well, in the received packet, the source IP address is src_ip = 192.168.1.1, the destination IP address is dst_ip = 192.168.10.2, and the protocol type is ip_type = 6.

  Accordingly, the OpenFlow switch 20 (processor 202) first compares the header of the received packet with the match condition of the high priority “3000” flow entry generated in FIG. 12B. Since the received packet satisfies the match condition of the flow entry with the priority “3000”, the received packet is output to the output port with the port number “2” and transferred according to the corresponding action. In this case, the number of packets processed up to that point = 2 packets (packet) is recorded as statistical information in the flow entry with the priority “3000” (broken line).

  In FIG. 12D, the OpenFlow switch 20 (processor 202) analyzes the next received packet, and the received packet matches one of the flow entries in the flow table in order of priority of the flow entry. Determine whether the condition is met. In this case, in the received packet, the source IP address is src_ip = 192.168.1.2, the destination IP address is dst_ip = 192.168.10.2, and the protocol type is ip_type = 6.

  Accordingly, the OpenFlow switch 20 (processor 202) first compares the header of the received packet with the match condition of the high priority “3000” flow entry generated in FIG. 12B. In this case, the received packet does not satisfy the match condition of the flow entry with the priority “3000”.

  Next, the OpenFlow switch 20 (processor 202) compares the header of the received packet and the like with the matching condition of the learning flow entry having the next higher or lower priority “1”. In this case, the learning action is executed because the received packet satisfies the matching condition of the learning type flow entry with the priority “1”.

  In the learning action, in the flow table of FIG. 12D, a new flow entry is generated that includes a new match condition based on the source address, destination address, and protocol type of the received packet. At this time, the source address, the destination address, and the protocol type of the received packet are extracted according to the learning match type of the learning action, and written in the flow table as a new flow entry match condition.

  In this case, the match conditions of the new flow entry are: match condition source IP address src_ip = 192.168.1.2 / 32, destination IP address src_ip = 192.168.10.2 / 32, and protocol type “ ip_type = 6 ″. In addition, the priority “3000” specified by the learning action and the remaining action “output = 2” of the learning type flow entry using the match condition as the operation condition are set in the new flow entry. .

  As a result, in the OpenFlow switch 20, another new flow entry including a matching condition of another type of received packet learned from another received packet based on the contents of the learning process is generated and set in the flow table. it can.

  Next, the OpenFlow switch 20 (processor 202) executes the remaining action “output = 2” other than the learning action. Therefore, the received packet is output to the output port with the port number “2” and transferred. The OpenFlow switch 20 (processor 202) further records the number of processed packets = 2 packets as a statistical information in the flow entry with the priority “1” (broken line). The OpenFlow switch 20 (processor 202) records the number of processed packets = 1 packet as a statistical information in a new flow entry with the priority “3000” (broken line).

  Next, the OpenFlow switch 20 (processor 202) sends the learning content notification message including the learning action identification information, the priority, and the learned matching condition (broken line) in the newly added flow entry to the OpenFlow controller. 10 to send. Next, the OpenFlow controller 10 (processor 102) extracts learning action identification information, priority, and learned match condition (broken line) from the received learning content notification message. Next, the OpenFlow controller 10 (processor 102) sends a flow entry including the extracted learned action identification information, priority, and learned match condition (broken line) to the flow for the OpenFlow switch 20 in the storage unit 104. Save to table.

  In this way, each time a different type of packet is received that satisfies the learning flow entry match condition according to the learning flow entry, a new high priority flow entry including a different matching condition is generated. Can be set in the flow table. Thereby, statistical information can be obtained for each different flow entry.

  Thereafter, the OpenFlow controller 10 (processor 102) transmits a statistical information request message for requesting statistical information of the flow table of the OpenFlow switch 20 to the OpenFlow switch 20. In response to the reception of the statistical information request message, the OpenFlow switch 20 (processor 202) extracts the statistical information of each flow entry in its own flow table, and sends the response message including the extracted statistical information to the OpenFlow controller. 10 to send. The OpenFlow controller 10 (processor 102) stores the received statistical information of each flow entry in the flow table for the OpenFlow switch 20 in the storage unit 104 and the statistical information corresponding to each flow entry.

  Also in the flow table of FIG. 12A, as in the case of FIG. 11A, one learning type flow entry is transmitted from the OpenFlow controller 10 to the OpenFlow switch 20 via the control line network 5. Accordingly, the time required for transmission of the learning type flow entry is short, and the bandwidth of the control circuit network 5 used for transmission of the learning type flow entry is small. Further, the number of learning type flow entries set in the flow table of the OpenFlow switch 20 is small. Therefore, the learning flow entry setting time in the OpenFlow switch 20 is also short.

  12B and 12D, the bandwidth used for transmitting the learning content notification message for flow table synchronization is the same as in FIG. 11B. That is, in the transmission of the learning content notification message for synchronizing the flow table from the OpenFlow switch 20 to the OpenFlow controller 10, different types of packets that satisfy the matching condition of the learning type flow entry are received. Occurs when. That is, the transmission of the learning content notification message occurs with time dispersion. Therefore, the bandwidth of the control circuit network 5 used for transmitting the learning content notification message can be small. Further, even if the transmission of the learning content notification message from the OpenFlow switch 20 to the OpenFlow controller 10 is delayed, the transfer service provided by the OpenFlow switch 20 is not affected.

  Next, generation and transmission of a flow entry including a learning algorithm by the OpenFlow controller 10 will be described.

  FIG. 13 shows an example of a flowchart of processing for generating and transmitting a flow entry including a learning algorithm, which is executed by the OpenFlow controller 10.

  Referring to FIG. 13, in step 502, the processor 102 (or the learning algorithm generation unit 1030 thereof) of the OpenFlow controller 10 generates a learning algorithm for generating a learning type flow entry according to the operation of the administrator. To do.

  In step 504, the processor 102 (or its flow entry generation unit 1024) generates a learning type flow entry or transfer control information including the generated learning algorithm as an action in accordance with the operation of the administrator. Here, the learning type flow entry includes, in addition to the learning action, other transfer processing actions such as output processing to a port. Next, the processor 102 (flow entry generation unit 1024) transmits the generated flow entry to the OpenFlow switch 20 via the control port 122 and the L2 switch 52.

  Next, reception and setting of a flow entry including a learning algorithm by the OpenFlow switch 20 will be described.

  FIG. 14 shows an example of a flowchart of processing executed by the OpenFlow switch 20 for receiving a flow entry including a learning algorithm and setting the flow entry in the flow table.

  Referring to FIG. 14, in step 512, the processor 202 (or its flow entry setting unit 2030) of the OpenFlow switch 20 sends a learning type flow entry from the OpenFlow controller 10 via the control port 222. Receive. Here, the learning type flow entry includes, in addition to a learning action, other transfer processing actions such as output processing to a port.

  In step 514, the processor 202 (flow entry setting unit 2030) sets the received flow entry in its own flow table. At that time, in addition to the learning action, for example, another transfer processing action such as output processing to a port is written as an action of the flow table.

  Next, processing of received packets based on the flow table by the OpenFlow switch 20 will be described. At that time, the match condition is learned from the received packet in accordance with the content of the learning action in the learning type flow entry, and a new flow entry including the learned match condition is added to the flow table.

  FIG. 15 shows an example of a flowchart executed by the OpenFlow switch 20 for processing a received packet based on a flow table. Here, the flow table includes a learning type flow entry including a learning action. The received packet is processed based on the learning type flow entry to learn the matching condition, and the flow entry including the matching condition is additionally set in the flow table.

  Referring to FIG. 15, in step 522, processor 202 (or its packet transfer unit 2026) receives a new packet from an input port in communication port 224.

  In step 524, the processor 202 (or its flow table reference unit 2024) refers to the flow table in the storage unit 204, and whether there is a match condition that matches (matches) the received packet in the flow entry of the flow table. Determine if. At that time, each flow entry in the flow table is referred to in the order of priority, and the match condition is determined. Here, as described above, the priority of the learning type flow entry is lower than the priority of the flow entry including the match condition generated by learning from the received packet by the learning action. Accordingly, the flow entry generated by the learning process is prioritized over the learning type flow entry, and it is determined whether or not the received packet matches the matching condition. If it is determined that there is no matching condition that matches the received packet, the procedure proceeds to step 526. If it is determined that there is a matching condition that matches the received packet, the procedure proceeds to step 528.

  In step 526, the processor 202 (or its packet-in message generator 2028) generates a packet-in message for inquiring how to process the received packet, and transmits it to the OpenFlow controller 10. The packet-in message is transmitted to the OpenFlow controller 10 via the control port 222 and the L2 switch 52. Transmission of a packet-in message is a normal form of known operation. Thereafter, the procedure exits the routine of FIG.

  In step 528, the processor 202 (or the received packet self-expanding unit 2032) determines whether the action of the flow entry including the matching condition matched with the received packet has a learning action. If it is determined that there is no learning action in the action of the flow entry, the procedure proceeds to step 530. If it is determined that the action of the flow entry has a learning action, the procedure proceeds to step 532.

  In step 530, the processor 202 (the packet transfer unit 2026 or the processing unit 2036) executes an action included in the flow entry. Here, the action is an action other than the learning action. For example, the processor 202 (packet transfer unit 2026) outputs a received packet to a designated output port and transfers it to another node on the network 7 as an action. Next, the processor 202 (the packet transfer unit 2026 or the processing unit 2036) writes new statistical information in the flow entry and updates the statistical information. Thereafter, the procedure exits the routine of FIG.

  In step 532, the processor 202 (received packet self-expanding unit 2032) self-expands the received packet received from the input port and learns a new match condition from the received packet. At this time, information defined by the learning match type of the learning action, for example, the transmission source address, the destination address, and the protocol type may be extracted from the header of the received packet as the learned matching condition.

  In step 534, the processor 202 (flow entry setting unit 2030) additionally sets a new flow entry having the learned match condition as an operation condition and including an action other than the learned action in the flow table of the storage unit 204. . Thereby, the processor 202 can generate a different flow entry for each type of received packet and set it in the flow table.

  In step 536, the processor 202 (the packet transfer unit 2026 or the processing unit 2036) executes an action other than the learning action. For example, the processor 202 (packet transfer unit 2026) outputs the received packet to a designated output port and transfers it to another node on the network 7 as another action. Next, the processor 202 (the packet transfer unit 2026 or the processing unit 2036) writes new statistical information in the flow entry, and updates the statistical information of the learning type flow entry and the newly added flow entry. In this way, new flow entry statistics are generated or updated.

  The action performed in steps 530 and 536 may be an action other than outputting the received packet to a specific output port. Such actions include, for example, rewriting a specified field, discarding a packet, specifying a queue to be placed, adding or deleting a header, and the like. Field rewriting includes, for example, rewriting the IP address or MAC address of the source and / or destination as the network identifier, rewriting the TCP or UDP port number of the source and / or destination, and / or rewriting the priority. Is included. The field rewriting includes, for example, VLAN tag rewriting and / or MPLS (Multi Protocol Label Switching) tag rewriting.

  In step 538, the processor 202 (or its learning content notification message generation unit 2034) transmits a learning content notification message including information on a new flow entry including the learned match condition to the open controller 10. At this time, the learning content notification message is transmitted to the open controller 10 via the control port 222 and the L2 switch 52. Thereafter, the procedure exits the routine of FIG.

  The learning content notification message transmitted in step 538 is added to the flow table for the OpenFlow switch 20 in the OpenFlow controller 10 as described later. As a result, the flow table in the OpenFlow switch 20 and the flow table for the OpenFlow switch 20 in the OpenFlow controller 10 match and can be synchronized. As described above, the bandwidth of the control line used for transmitting the learning content notification message for such flow table synchronization is small.

  Thus, the OpenFlow switch 20 can process the received packet according to the received packet, generate a different new flow entry according to the type of the received packet, and set it in the flow table. Further, the OpenFlow switch 20 can collect statistical information regarding received packets processed for different flow entries.

  Next, processing for updating the flow table for the OpenFlow switch 20 in the OpenFlow controller 10 by the OpenFlow controller 10 will be described.

  FIG. 16 shows an example of a flowchart of processing for updating the flow table for the OpenFlow switch 20 in the storage unit 104 of the OpenFlow controller 10 executed by the OpenFlow controller 10.

  In step 542, the processor 102 (or its flow table synchronization processing unit 1028) receives the learning content notification message from the OpenFlow switch 20 via the control port 122 via the L2 switch 52.

  In step 544, the processor 102 (flow table synchronization processing unit 1028) extracts the flow entry information from the learning content notification message and stores it in the flow table for the open flow switch 20 in the storage unit 104. Thereby, the flow tables of both the OpenFlow switch 20 and the OpenFlow controller 10 are set or synchronized to match each other. Even if the transmission of the learning content notification message from the OpenFlow switch 20 to the OpenFlow controller 10 is delayed, only the synchronization in the OpenFlow controller 10 is delayed, and the transfer service in the OpenFlow switch 20 is performed. It will not be affected.

  Next, collection of flow table statistical information in the OpenFlow switch 20 by the OpenFlow controller 10 will be described.

  FIG. 17 shows an example of a flowchart of processing for collecting statistical information from the OpenFlow switch 20 executed by the OpenFlow controller 10.

  Referring to FIG. 17, in step 552, the processor 102 (or its statistical information collection unit 1026) of the OpenFlow controller 10 transmits a statistical information request message requesting statistical information to the OpenFlow switch 20. The statistical information request message is transmitted to the OpenFlow switch 20 via the L2 switch 52 via the control port 222.

  In step 554, the processor 202 (flow table reference unit 2024 or processing unit 2036) of the OpenFlow switch 20 receives the statistical information request message from the OpenFlow controller 10.

  In step 556, the processor 202 (flow table reference unit 2024) extracts statistical information of each flow entry in the flow table. As an alternative, the information of all flow entries in the flow table may be retrieved together with the respective statistical information.

  In step 558, the processor 202 (flow table reference unit 2024 or processing unit 2036) transmits a response message including the extracted statistical information to the OpenFlow controller 10.

  In Step 560, the processor 102 (statistical information collecting unit 1026) of the OpenFlow controller 10 receives a response message including statistical information from the OpenFlow switch 20.

  In step 562, the processor 102 (statistical information collecting unit 1026) extracts the statistical information from the response message and stores it in each flow entry of the flow table for the open flow switch 20 in the storage unit 104.

  In this way, the OpenFlow controller 10 can collect statistical information about each flow entry generated and set by learning in the OpenFlow switch 20 from the OpenFlow switch 20.

  In the above description, when the matching condition in the learning type flow entry and each flow entry generated by the learning process includes a match of an IP address such as a source and / or destination IP address as a network identifier Explained. On the other hand, the embodiment is not limited to this, and other network such as a MAC address, a VLAN (Virtual LAN) tag, or the like may be used as a matching condition in the learning type flow entry and the flow entry generated by the learning process. It can also be applied when an identifier match is included.

  Thus, according to the embodiment, the OpenFlow controller 10 sets a learning type flow entry in the OpenFlow switch 20. Accordingly, the OpenFlow controller 10 can set a plurality of flow entries in the flow table by learning processing based on one learning type flow entry corresponding to the type of received packet.

  Further, according to the embodiment, the OpenFlow controller 10 sets individual flow entries corresponding to the type of received packet in the OpenFlow switch 20 in the flow table of the OpenFlow switch 20 in a short time. be able to. Thereby, the OpenFlow controller 10 can further shorten the interruption time of the packet transfer service that may occur when the OpenFlow switch fails during operation of the OpenFlow switch.

  In addition, according to the embodiment, the communication carrier managing the OpenFlow controller 10 and the OpenFlow switch 20 can reduce the bandwidth of the control line secured between the OpenFlow controller 10 and the OpenFlow switch 20. Can be small. Thereby, the communication carrier can reduce the maintenance cost of the control line between the OpenFlow controller 10 and the OpenFlow switch 20.

  All examples and conditional expressions given here are intended to help the reader understand the inventions and concepts that have contributed to the promotion of technology, such examples and It is understood that the present invention is not limited to the conditions, and that the organization of such examples in the specification is not related to the superiority or inferiority of the present invention. Although embodiments of the present invention have been described in detail, it will be understood that various changes, substitutions and variations can be made thereto without departing from the spirit and scope of the invention.

Regarding the embodiment including the above examples, the following additional notes are further disclosed.
(Additional remark 1) The transfer control apparatus which produces | generates and transmits the 1st transfer control information containing the learning process and transfer process performed on the 1st conditions containing the information showing a some network identifier,
When the first transfer control information is received from the transfer control device and stored in a storage unit, and the received packet is referred to the storage unit and the packet satisfies the first condition, the learning process is performed. A transfer device that generates second transfer control information including the transfer process executed under a second condition including a network identifier of the packet, stores the second transfer control information in the storage unit, and processes the packet according to the transfer process; ,
Including transfer system.
(Supplementary Note 2) The transfer device refers to the storage unit for another received packet and processes the other packet according to the transfer process when the other packet satisfies the second condition. The transfer system according to appendix 1, wherein there is a transfer system.
(Supplementary note 3) The transfer system according to Supplementary note 1 or 2, wherein the second transfer control information has higher reference priority than the first transfer control information.
(Additional remark 4) The said transfer apparatus preserve | saves the statistical information regarding the packet processed according to said 1st transfer control information and said 2nd transfer control information, Any of Additional remark 1 thru | or 3 characterized by the above-mentioned. The transfer system described in.
(Additional remark 5) The said transfer apparatus refers to the said memory | storage part about the other packet received, and when the said another packet does not satisfy | fill the said 2nd conditions, but according to the said learning process, Third transfer control information including the transfer process executed under a third condition including the network identifier of the another packet is generated and stored in the storage unit, and the other packet is transferred according to the transfer process The transfer system according to any one of appendices 1 to 4, wherein the transfer system is a device.
(Additional remark 6) The said transfer apparatus preserve | saves the statistical information regarding the packet processed according to said 1st transfer control information, said 2nd transfer control information, and said 3rd transfer control information, It is characterized by the above-mentioned. The transfer system according to appendix 5.
(Supplementary note 7) The transfer system according to any one of Supplementary notes 1 to 6, wherein the transfer process is to output the received packet to a designated output unit.
(Appendix 8) The transfer device transmits the generated second transfer control information to the transfer control device,
8. The transfer system according to any one of appendices 1 to 7, wherein the transfer control device receives the second transfer control information and stores the second transfer control information in a storage unit of the transfer control device.
(Additional remark 9) The memory | storage part which memorize | stores the 1st transfer control information containing the learning process and transfer process which are received on the 1st condition containing the information showing a some network identifier received from the transfer control apparatus,
The received packet is referred to the storage unit, and when the packet satisfies the first condition, the transfer process is executed including the transfer process executed under the second condition including the network identifier of the packet according to the learning process. Processing unit that generates and stores the transfer control information of 2 in the storage unit, and processes the packet according to the transfer process;
Including transfer device.
(Supplementary Note 10) The transfer device is
The first transfer control information including the learning process and the transfer process executed under the first condition including information representing a plurality of network identifiers received from the transfer control apparatus is stored, and the received packet is referred to the storage unit When the packet satisfies the first condition, the second transfer control information including the transfer process executed under the second condition including the network identifier of the packet is generated according to the learning process, and A transfer method for storing in a storage unit and executing processing for processing the packet according to the transfer processing.
(Supplementary Note 11) The transfer device is
The first transfer control information including the learning process and the transfer process executed under the first condition including information representing a plurality of network identifiers received from the transfer control apparatus is stored, and the received packet is referred to the storage unit When the packet satisfies the first condition, the second transfer control information including the transfer process executed under the second condition including the network identifier of the packet is generated and stored according to the learning process. A program for executing processing for storing the packet and processing the packet according to the transfer processing.

2 Transfer System 5 Control Line Network 52 L2 Switch 7 Network 10 Open Flow Controller 102 Processor 104 Storage Unit 20, 22, 24 Open Flow Switch 202 Processor 204 Storage Unit

Claims (8)

A transfer control device for generating and transmitting first transfer control information including learning processing and transfer processing executed under a first condition including information representing a plurality of network identifiers;
When the first transfer control information is received from the transfer control device and stored in a storage unit, and the received packet is referred to the storage unit and the packet satisfies the first condition, the learning process is performed. A transfer device that generates second transfer control information including the transfer process executed under a second condition including a network identifier of the packet, stores the second transfer control information in the storage unit, and processes the packet according to the transfer process; ,
Including transfer system.   The transfer device refers to the storage unit for another received packet, and processes the other packet according to the transfer process when the other packet satisfies the second condition. The transfer system according to claim 1.   The transfer system according to claim 1 or 2, wherein the second transfer control information has higher reference priority than the first transfer control information.   The transfer device refers to the storage unit for another received packet, and when the other packet does not satisfy the second condition and satisfies the first condition, the second packet is transmitted according to the learning process. Generating third transfer control information including the transfer process executed under the third condition including the network identifier of the network identifier, storing the third transfer control information in the storage unit, and transferring the other packet according to the transfer process The transfer system according to claim 1, wherein: The transfer device transmits the generated second transfer control information to the transfer control device,
The transfer system according to any one of claims 1 to 7, wherein the transfer control device receives the second transfer control information and stores it in a storage unit of the transfer control device. A storage unit that stores first transfer control information including learning processing and transfer processing executed under a first condition including information representing a plurality of network identifiers received from the transfer control device;
The received packet is referred to the storage unit, and when the packet satisfies the first condition, the transfer process is executed including the transfer process executed under the second condition including the network identifier of the packet according to the learning process. Processing unit that generates and stores the transfer control information of 2 in the storage unit, and processes the packet according to the transfer process;
Including transfer device. The transfer device
Storing first transfer control information including learning processing and transfer processing executed under a first condition including information representing a plurality of network identifiers received from the transfer control device;
The received packet is referred to the storage unit, and when the packet satisfies the first condition, the transfer process is executed including the transfer process executed under the second condition including the network identifier of the packet according to the learning process. A transfer method for generating transfer control information 2 and storing the transfer control information in the storage unit and processing the packet according to the transfer process. The transfer device
Storing first transfer control information including learning processing and transfer processing executed under a first condition including information representing a plurality of network identifiers received from the transfer control device;
A second process that includes the forwarding process that is executed under a second condition including a network identifier of the packet according to the learning process when the packet satisfies the first condition with reference to the storage unit for the received packet; A program for executing the process of generating the transfer control information and storing the transfer control information in the storage unit and processing the packet according to the transfer process.

Images