JP2014150349A - Communication apparatus and communication apparatus control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To achieve power saving of a communication apparatus having a plurality of ports which compose a link aggregation group.SOLUTION: A communication apparatus comprises: a plurality of physical ports; a plurality of processing circuits for processing packets of the plurality of physical ports; a first control part for controlling power consumption of the plurality of processing circuits based on traffics of the plurality of physical ports; first information for controlling states of the plurality of physical ports in power consumption control of the plurality of processing circuits by the first control part; and a second control part. The second control part aggregates the plurality of physical ports into one group and uses the group as one logical port; with reference to the first information, selects a plurality of using physical ports from the group, for distributing transmission packets based on the states of the plurality of physical ports; and selects a physical port from the plurality of using physical ports, for transmitting the packets in transmission of packets through the logical ports.

Description

  The present invention relates to a communication device and a communication device control method.

  With the expansion of the use of information communication via the network, the number of communication devices such as routers and switches has increased, and the amount of traffic in the network has increased. Under such circumstances, Patent Document 1 discloses a power saving technique for a communication device.

  There are several methods for reducing the power consumption of the communication device, one of which is to devise the operation of the communication device. For example, as a method for reducing the power consumption of Ethernet (registered trademark), Energy Efficient Ethernet (EEE) defined in IEEE 802.3az is known.

  A communication device in EEE shifts a link (port) in which no traffic is generated to a power saving state called a low power idle (LPI) state. The communication device transmits and receives only a signal (referred to as a refresh signal) for holding the link periodically during the LPI state. This reduces power consumption in the physical layer of the network.

  On the other hand, link aggregation, which is a technology for making communication lines redundant, is known. Link aggregation can broaden the communication bandwidth by logically combining a plurality of links (ports) into one link. Furthermore, in the link aggregation, even if a failure occurs in one link (port), communication can be continued using another link, so that fault tolerance can be improved.

JP 2010-213259 A

  In the Ethernet Interface Card (NIC), the Ethernet control (EEE) reduces the power supplied to the control circuit that processes data transmitted / received via the port during the LPI state, thereby reducing the power consumption of the communication device. Can be reduced. That is, the control circuit is in a power saving state during the LPI state.

  When the packet arrives at the NIC during the LPI state, the communication apparatus needs to change the control circuit from the power saving state to the active state capable of data processing. The state transition between the power saving state and the active state reduces the power saving effect due to its overhead.

  In link aggregation, a load caused by traffic can be distributed by grouping a plurality of ports (links) and appropriately selecting ports (links) for transmitting packets. When a communication device uses link power saving technology together with link aggregation, the communication device controls them independently, that is, if a port (link) used for packet transmission is selected without referring to the port status, The effect is greatly reduced.

  According to one embodiment of the present invention, a plurality of physical ports, a plurality of processing circuits that process packets of the plurality of physical ports, and power consumption of the plurality of processing circuits are controlled based on traffic of the plurality of physical ports. The communication device includes: a first control unit; first information for managing states of the plurality of physical ports in power consumption control of the plurality of processing circuits by the first control unit; and a second control unit. . The second control unit combines the plurality of physical ports into one group, uses the group as one logical port, and refers to the first information based on the state of the plurality of physical ports. A plurality of used physical ports for distributing transmission packets are selected from the group, and a physical port for transmitting the packets is selected from the plurality of used physical ports in packet transmission via the logical port.

  According to one aspect of the present invention, it is possible to realize power saving of a communication apparatus having a plurality of ports that constitute a link aggregation group.

It is a block diagram which shows typically the structural example of the packet relay apparatus in this embodiment. In this embodiment, it is a figure which shows typically the power consumption of the control circuit in the port state in each LPI control, and each port state. It is a figure which shows the structural example of the port state management table of this embodiment. It is a figure which shows the structural example of the LA state management table of this embodiment. In this embodiment, it is a figure which shows the example of an item which can be user-set in the structure of LA group. It is a figure which shows the structural example of the LA setting management table of this embodiment. It is a flowchart which shows the process example of the notification of a port state change by the LPI control part in this embodiment. It is a flowchart which shows the example of the reselection process of the LA use port by the layer 2 control part of this embodiment. 9 is a flowchart illustrating details of a port selection step (S209) in the performance-oriented mode in the flowchart of FIG. It is a block diagram which shows typically the other structural example of the packet relay apparatus in this embodiment.

  Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. The following description and the accompanying drawings show specific embodiments and implementation examples according to the principle of the present invention, but these are for the understanding of the present invention and should not be interpreted in a limited manner. It is not intended for use.

  The communication apparatus of this embodiment has a function of reducing power consumption of a physical port (physical link) and a link aggregation function that uses a plurality of physical links (physical ports) as one logical link (logical port). Yes. A plurality of grouped physical ports (physical links) are also referred to as LA (Link Aggregation) groups.

  In the present embodiment, an example will be described in which a port (link) used for data transmission is selected from ports (links) constituting a ring aggregation group in consideration of the state of the port (link) in power consumption control.

  The example described below is a technology for reducing the power consumption of Ethernet (registered trademark), and employs Energy Efficient Ethernet (EEE) defined in IEEE 802.3az. A communication device in EEE shifts a port (link) in which no traffic is generated to a power saving state called a Low Power Idle (LPI) state (described later with reference to FIG. 2).

  In the LPI state, the communication devices on both sides of the link can reduce the power consumption by stopping a part of the functions (circuits). EEE is a power-saving technology in Ethernet, which is a data link layer, and a communication apparatus stops power supply to a circuit (chip) that processes a protocol of the data link layer to reduce power consumption.

  The communication apparatus according to the present embodiment refers to the state of the port in the LA group (operation state in LPI control) in data communication by the LA group, and uses it for data transmission for load distribution based on the state. Select a port. The communication device sequentially selects a port for transmitting a packet from the selected ports according to a predetermined algorithm.

  In this way, effective power saving is realized for a communication apparatus having physical ports constituting a link aggregation group by selecting a port that distributes transmission packets from the LA group based on the port state in power consumption control. .

  FIG. 1 is a block diagram schematically illustrating a configuration example of a packet relay device 10A, which is an example of a communication device. The packet relay device 10A is, for example, a layer 2 switch or a layer 3 switch. This embodiment has a feature in the function in the data link layer (layer 2), and FIG. 1 shows a configuration for explaining the function.

  The packet relay apparatus 10 </ b> A includes a CPU 101 that is a processor, a memory 102, and a packet processing unit 103, which are communicably connected via a bus 104. The memory 102 stores a plurality of programs executed by the CPU 101, and a layer 2 control program 121 is illustrated in FIG. The CPU 101 performs various processes by operating in accordance with programs stored in the memory 102 including the layer 2 control program 121.

  The packet relay device 10A includes a plurality of ports (physical ports) 105 that perform conversion between electrical signals and data. In the example of FIG. 1, the ports 105A to 105D are specifically illustrated. Port 105 represents any one or more ports.

  The packet processing unit 103 includes a layer 2 control unit 131, a packet buffer 132, and a port distribution control unit 133. The function of the packet processing unit 103 can be realized by a processor that operates according to a program, hardware designed by an integrated circuit, or a combination thereof.

  The layer 2 control unit 131 includes a port selection unit 315, an LA (Link Aggregation) state management table 137, and an LA setting management table 138. As will be described later, based on the state of the port 105, the port selection unit 315 reselects a plurality of used physical ports (LA used ports in the following description) that distribute transmission packets from the LA group. The port selection unit 315 further selects one physical port that transmits the packet from the selected LA use port in the packet transmission via the LA group.

  The port allocation control unit 133 includes an LPI (Low Power Idle) control unit 134 and a MAC (Media Access Control) control unit 136. The MAC control unit 136 processes the packet of the port 105.

  The LPI control unit 134 controls power supply to each of the MAC control units 136 based on the traffic of the port 105. The LPI control unit 134 includes a port state change notification unit 345 and a port state management table 135. The port state management table 135 is a table for managing the state of the port 105 (the state of the port in the power supply control of the MAC control unit 136). When the state of the port 105 changes, the port state change notifying unit 345 displays the layer state. 2 Notify the control unit 131.

  The packet relay device 10A is connected to another packet relay device 10B via a plurality of network lines 106. In the example of FIG. 1, the network lines 106A to 106D to which the ports 105A to 105D are connected are specifically illustrated. The network line 106 indicates any one or a plurality of network lines. In this example, a plurality of ports including the ports 105A to 105D constitute an LA group.

  The MAC control unit 136 is a circuit that processes a protocol of the MAC layer. In this example, layer 2 and layer 3 data units are referred to as packets. The MAC control unit 136 is connected to the plurality of ports 105 and processes packets transmitted and received through them. For example, the MAC control unit 136 gives layer 2 information necessary for communication with the opposite device to the packet, cuts out a layer 3 packet, and controls packet transmission / reception.

  The MAC control unit 136 includes a plurality of processing circuits (for example, chips), and one processing circuit processes a packet of one or a plurality of assigned ports 105. The LPI control unit 134 can control power feeding for each processing circuit in the MAC control unit 136.

  In the following description, one processing circuit is responsible for one port 105, and the LPI control unit 134 controls whether or not to supply power necessary for the operation of each processing circuit in the MAC control unit 136 according to the EEE protocol. To do. When the supply of power necessary for the operation is stopped, the processing circuit cannot process the packet. For example, the LPI control unit 134 designates a port and instructs the MAC control unit 136 to perform power supply control, and the MAC control unit 136 stops power supply to the processing circuit in charge of the designated port.

  As described above, the LPI control unit 134 changes the state of one link (port) by controlling the power supply of the processing circuit of the MAC control unit 136. FIG. 2 is a diagram for explaining a state of a link (port) by LPI control.

  FIG. 2 schematically shows the state change of the link (port) and the power consumption in each state. In the lower graph of FIG. 2, the X-axis indicates time, and the Y-axis indicates link power consumption. As described above, here, the power consumption (supply power) of the corresponding processing circuit in the MAC control unit 136 is shown. When the processing circuit in the MAC control unit 136 takes charge of a plurality of ports, the plurality of ports exhibit the same state change.

  In LPI control, five states of links (ports) are defined. They are an active state, a sleep state, a quiet state, a refresh state, and a wake state. In the following description, these are called port states. The LPI state is a power saving state and is a quiet state or a refresh state, and one period of the LPI state is a period of a continuous quiet state or a refresh state.

  The active state is a normal power state. In this state, the MAC control unit 136 is supplied with power necessary for operation. In the active state, the link (port) can transmit and receive traffic data that requires processing at the MAC layer by the MAC control unit 136.

  In the quiet state, power supply to the corresponding processing circuit in the MAC control unit 136 is stopped, and power necessary for operation for packet transmission / reception is not supplied. The sleep state is a transition state from the active state to the quiet state. The wake state is a transition state from the quiet state to the active state. In the sleep state and the wake state, the link (port) cannot transmit traffic data. In the sleep state and the wake state, power is supplied to the processing circuit.

  When the port state enters the quiet state, the LPI control units 134 at both ends of the link periodically transmit and receive a refresh control message (refresh signal). This state is a refresh state. This avoids the link being quiet for a long time. Since this control message is processed by skipping the MAC control unit 136, power supply to the corresponding processing circuit in the MAC control unit 136 remains stopped.

  The MAC controller 136 may be configured to provide a limited LPI state. In the limited LPI state, the MAC control unit 136 shifts only the circuit on the transmission side to the power saving state and keeps the circuit receiving the packet from the opposite side activated. For example, when the LPI control unit 134 does not receive the consent to the control message indicating the shift to the LPI state from the opposite packet relay apparatus 10B, the LPI control unit 134 causes the MAC control unit 136 to shift to the limited LPI state of the processing circuit. Instruct.

  Here, the control of the port state by the LPI control unit 134 will be described. FIG. 3 shows a configuration example of the port state management table 135 used by the LPI control unit 134. The port state management table 135 includes a column 351 that stores a port number that is an identifier of a port, and a column 352 that stores a value indicating the port state. The LPI control unit 134 can know the state of each port with reference to the port state management table 135.

  Although not shown in FIG. 3, the port state includes a state where a failure has occurred in addition to a state based on LPI control. This state is also managed in the port state management table 135. The port distribution control unit 133 includes a failure detection unit. When a failure such as a cable disconnection or a port failure is detected, the port distribution control unit 133 updates the port state management table 135 and notifies the layer 2 control unit 131 of the update.

  When the port state of any port is changed, the LPI control unit 134 (specifically, the port state change notification unit 345) updates the port state management table 135 to reflect the change. Furthermore, as will be described later, the LPI control unit 134 (specifically, the port state change notification unit 345) notifies the layer 2 control unit 131 when the state of any port changes.

  The LPI control unit 134 controls the port state based on the traffic data at the port. For example, the LPI control unit 134 changes the port state from the active state to the LPI state when there is no transmission of traffic data at the port for a predetermined period. In addition, when receiving a control message related to LPI, the LPI control unit 134 controls the port state according to the message.

  The LPI control unit 134 stops the corresponding processing circuit of the MAC control unit 136 to shift the port 105 to the LPI state, and activates the corresponding processing circuit of the MAC control unit 136 to return the port 105 to the active state.

  First, a method in which the LPI control unit 134 changes the port state from the active state to the LPI state will be described. This state change is determined by the LPI control unit 134 according to a predetermined condition, or is instructed by a control message indicating a transition from the outside to the LPI state.

  In this state change, the LPI control unit 134 instructs the MAC control unit 136 to stop power supply to the corresponding processing circuit. In response to the message, the MAC control unit 136 stops the processing circuit for transmitting / receiving the traffic data (traffic packet) of the port and puts it in the power saving state.

  The LPI control unit 134 changes the port state from the active state to the LPI state when there is no transmission of traffic data at the port for a predetermined period. When it is detected that a signal including traffic is not sent from the upstream of the packet buffer 132 or the layer 2 control unit 131 for a predetermined time, the LPI control unit 134 determines the transition from the active state to the LPI state, and the MAC control unit The corresponding processing circuit at 136 is stopped.

  The LPI control unit 134 transmits a control message indicating the transition to the LPI state from the port to the opposite packet relay apparatus 10B. The LPI control unit 134 updates the port state management table 135 and notifies the layer 2 control unit 131 of changes in the port state.

  Next, state transition by the control message will be described. The LPI control unit 134 sends a control message for instructing the transition to the LPI state from an element in the packet relay device 10A, for example, from the layer 2 control unit 131 or the CPU 101, or from the opposite packet relay device 10B via the network line 106. When received, the state of the corresponding port is changed from the active state to the LPI state. The port distribution control unit 133 determines the data type received from the port 105 and transfers the control message to the LPI control unit 134 without passing through the MAC control unit 136.

  When receiving the control message indicating the transition to the LPI state, the LPI control unit 134 instructs the MAC control unit 136 to stop the processing circuit (circuit for packet transmission / reception) corresponding to the target port. The MAC control unit 136 stops the processing circuit according to the instruction. The LPI control unit 134 transmits an agreement message via the port 105 and the network line 106 in response to the control message indicating the transition from the opposite packet relay apparatus 10B to the LPI state. When the port state is changed, the LPI control unit 134 updates the port state management table 135 and notifies the layer 2 control unit 131 of the port state change.

  Next, a method in which the LPI control unit 134 returns the port state to the active state will be described. When receiving the control message indicating the return to the active state, the LPI control unit 134 shifts the state of the designated port to the active state. The control message indicating the return to the active state is transmitted from another element in the packet relay device 10A, for example, the layer 2 control unit 131 or the CPU 101, and the LPI control unit 134 receives the link from the opposite packet relay device 10B of the link. Receive via network line 106 and port 105.

  When the LPI control unit 134 receives a control message indicating the return to the active state designating the port from another element in the apparatus, for example, the layer 2 control unit 131, the LPI control unit 134 passes the port 105 and the corresponding network line 106. Then, a control message indicating return to the active state is transmitted to the opposite packet relay apparatus 10B. Further, the LPI control unit 134 instructs the MAC control unit 136 to start the processing circuit corresponding to the designated port.

  When the LPI control unit 134 receives a response from the MAC control unit 136 and the opposite packet relay apparatus 10B, the LPI control unit 134 updates the port state management table 135 and further transmits a control message indicating that the designated port has been returned to the active state. For example, the layer 2 control unit 131 is notified. When the control message transmission source is not the layer 2 control unit 131, the LPI control unit 134 further notifies the layer 2 control unit 131 of the change of the port state.

  When the LPI control unit 134 receives a control message indicating the return to the active state from the opposite packet relay apparatus 10B, the LPI control unit 134 instructs the MAC control unit 136 to start the processing circuit corresponding to the designated port. When receiving a response to the instruction from the MAC control unit 136, the LPI control unit 134 transmits a response to the control message to the opposite packet relay apparatus 10B. The LPI control unit 134 updates the port state management table 135 and notifies the layer 2 control unit 131 of the change of the port state.

  Next, traffic packet processing different from the LPI control message will be described. The traffic packet is processed by the MAC control unit 136. In this example, the layer 2 control unit 131 uses link aggregation. In the link aggregation, a plurality of physical links (physical ports) are grouped and used as one logical link (logical port). With link aggregation, it is possible to widen the communication band and improve fault tolerance.

  FIG. 4 shows a configuration example of the LA state management table 137 used by the layer 2 control unit 131 in the link aggregation of this embodiment. The LA state management table 137 includes an LA number column 371, a port number column 372, a port state column 373, an LA state column 374, a LACP (Link Aggregation Control Protocol) state column 375, and a port priority column 376.

  The LA number column 371 stores an LA number, which is an identifier of an LA group (a group of physical ports grouped by link aggregation) to which the port (physical port) 105 belongs. The port number column 372 stores a port number that is an identifier of the port 105. The port status column 373 stores a value indicating the status of the port 105 in LPI control.

  The LA state column 374 stores a value indicating that the port is in operation (ACT in FIG. 4) or a value indicating that the port is in standby (SBY in FIG. 4) in link aggregation. Hereinafter, a port whose LA state is “ACT” is called an ACT port, and a port whose LA state is SBY is called an SBY port.

  The LACP status column 375 stores a value indicating whether a link is established or not established by LACP. A traffic packet can be communicated only when a link by LACP is established. As will be described later, in this example, the packet relay apparatus 10A supports a plurality of LA modes, one of which is the LACP mode. The entry of the LA for which another mode is selected stores a NULL value in the LACP status column 375, for example.

  The port priority column 376 stores the priority of the port designated by the user. The layer 2 control unit 131 selects an SBY port (ACT port) based on the port priority. The layer 2 control program 121 passes the port priority to the layer 2 control unit 131 when defining the LA group.

  For example, when any of the port state, the LA state, or the LACP state of the port changes, the layer 2 control unit 131 updates the LA state management table 137 to reflect the change. As described above, when the port state is changed, the LPI control unit 134 notifies the layer 2 control unit 131 of this change. The layer 2 control unit 131 updates the LA state management table 137 according to the notification.

  The LA state will be described. In this example, the LA group can include an SBY port (SBY link) in addition to the ACT port. An SBY port is prepared in advance in the LA group. When a failure is detected in the link of the ACT port, the layer 2 control unit 131 switches the port that can be used for traffic packet communication from the port of the failed link to the SBY port. This maintains the number of links available in the LA group.

  The detected failure includes, for example, a cable disconnection, a hardware failure of a port in the own device 10A or the opposite device 10B, and a link failure detected by LACP. The port allocation control unit 133 detects a hardware failure, updates the port state management table 135, and notifies the layer 2 control unit 131 of this. In response to the port failure notification from the port allocation control unit 133, the layer 2 control unit 131 updates the LA state management table 137. As will be described later, a failure (link not established) detected by LACP is also reflected in the LA state management table 137.

  In this example, the user can set the configuration of each LA group in the packet relay apparatus 10A using the management computer. The layer 2 control program 121 stores the input information in the memory 102 and the memory in the layer 2 control unit 131. FIG. 5 shows an example of user setting items for the configuration of the LA group.

  As shown in FIG. 5, for example, the user can select the LA number of the LA group, the identifiers (here, port numbers) of a plurality of ports constituting the LA group, the priority of each port, the maximum number of operating links, the LA mode and the load. The port mode for distribution can be specified. The LA configuration user settings are reflected in the management tables 135, 137, and 138 of the packet relay apparatus 10A by the layer 2 control program 121.

  FIG. 6 shows a configuration example of the LA setting management table 138. The LA setting management table 138 has an LA number column 381, an LA mode column 382, a maximum number of operating links column 383, and a load distribution port mode column 384. The LA setting management table 138 stores information on the LA mode, the maximum number of operational links, and the load balancing port mode designated by the user for each LA group.

  In the initial setting, the number of ports that belong to the LA group exceeds the designated maximum operation link number is set as the SBY port. The SBY port is selected from the LA group based on the port number and the port priority specified by the user. The maximum number of operating links matches the number of usable ports, that is, the number of ports in which the LA state is “ACT” and no failure has occurred.

  When a failure occurs in an ACT port (ACT link) (when the port state or the LACP state is changed), the layer 2 control unit 131 sets a port whose LA state is “SBY” in the same LA group to the LA state management table. Select from 137 and change LA state to "ACT". For example, the port with the highest port priority is selected.

  The layer 2 control unit 131 updates the entry value of the port whose LA state has been changed in the LA state management table 137. In this example, the LA state of the entry of the selected SBY port is changed from “SBY” to “ACT”. In this example, the value of the port status column 373 or the LACP status column 375 is updated in the entry of the port where the failure has occurred, and the LA status is maintained at “ACT”.

  As shown in FIG. 5, in this example, the user can designate a link aggregation mode (LA mode). The packet relay device 10A supports the LACP mode and the static mode as the LA mode. In this example, the user can select the link aggregation mode for each LA group.

  The static mode starts operation when a link designated as an LA group is linked up (communication enabled in physical connection) without performing negotiation.

  The LACP mode is a link aggregation using LACP (specified in IEEE 802.3ad). The LACP mode performs negotiation between devices in accordance with LACP in grouping ports after link-up.

  For example, the layer 2 control programs 121 of the packet relay apparatuses 10A and 10B at both ends of the link negotiate by transmitting and receiving control messages (referred to as LACP packets). When the LACP packet is normally exchanged, a logical link (data link layer link) between the ports is established. The layer 2 control program 121 periodically exchanges LACP packets even after the link is established, and maintains the link. By using LACP, it is possible to improve the accuracy of consistency check between devices, link normality confirmation and failure detection.

  If the layer 2 control program 121 does not receive the LACP packet within the specified time on the link in the LA group, the layer 2 control program 121 notifies the layer 2 control unit 131 of the reception. In the LA state management table 137, the layer 2 control unit 131 changes the value of the LACP state column 375 of the entry of the link from “established” to “not established”. This is one of the failures of the ACT link (port).

  The layer 2 control unit 131 switches the operating port (link) from the failed link (port) to the SBY port (link). In the LA state management table 137, the layer 2 control unit 131 maintains the value of the LA state column 374 of the failed port entry as “ACT”, and sets the value of the LA state column 374 of the entry of the switching destination port as “ACT”. Change to

  The layer 2 control program 121 does not negotiate on the SBY link that is not used for actual packet communication, and its LACP state is “unestablished”. The layer 2 control program 121 performs negotiation according to LACP in the new ACT link, and establishes the link.

  The layer 2 control unit 131 distributes the packet to a plurality of ports (links) by transmitting the packet by one port selected from the ACT ports in transmission of the traffic packet using the LA group as one logical link. To do. The layer 2 control unit 131 according to the present embodiment selects ports to be used for packet transmission and to distribute traffic packets from the ACT ports based on their port states.

  First, an example of the flow of traffic packets in the packet relay apparatus 10 and the processing of each unit in the transfer of traffic packets (reception and transmission of packets from outside) will be described.

  The port distribution control unit 133 sends a traffic packet received at the port 105 from the outside via the network line 106 to the MAC control unit 136. The port distribution control unit 133 further notifies the LPI control unit 134 of the arrival of the packet.

  The processing circuit in charge of the port 105 in the MAC control unit 136 performs processing such as encoding on the received traffic packet, and forwards it to the layer 2 control unit 131. The layer 2 control unit 131 temporarily stores the traffic packet received from the MAC control unit 136 in the packet buffer 132.

  The traffic packet includes information such as a destination MAC address, a source MAC address, a destination IP address, a source IP address, a destination TCP port number, a source TCP port number, and data. In the following description, the destination address refers to the destination MAC address. A destination IP address may be used.

  The layer 2 control unit 131 refers to the route table and searches for the output destination using the destination address included in the traffic packet as a key. The route table associates the destination address of the received traffic packet with its output destination port. One output destination port is one logical port corresponding to one physical port 105 or LA group. For example, the physical port 105 is indicated by a port number, and the logical port is indicated by an LA number.

  The route table may be set by the user using a management computer, for example, or the layer 2 control program 121 may automatically set the output destination of the destination address by MAC address learning processing.

  When the output destination of the traffic packet is unknown, the layer 2 control program 121 performs destination resolution of the unknown destination. In addition, the layer 2 control program 121 performs update of information in the packet relay device 10A, aging processing, and the like.

  When the output destination port is the physical port 105, the layer 2 control unit 131 sends a traffic packet from the packet buffer 132 to the MAC control unit 136 by designating the port number. The MAC control unit 136 performs processing such as encoding on the received traffic packet by the processing circuit in charge of the output destination port 105, and sends the processed traffic packet to the port 105.

  When the port state of the output destination port 105 is not active, the layer 2 control unit 131 changes the port state of the port 105 to the active state to the LPI control unit 134 before sending the traffic packet to the MAC control unit 136. Send a control message to tell

  For example, the LA state management table 137 stores information on all ports 105 including ports that do not belong to any LA group, or other tables for managing the port states of all ports 105 are prepared. The layer 2 control unit 131 can know the port state of the port by referring to these.

  The layer 2 control unit 131 waits for the port 105 (corresponding processing circuit of the MAC control unit 136) to become active. When the layer 2 control unit 131 receives a notification from the LPI control unit 134 that the designated port 105 is in an active state, the layer 2 control unit 131 transmits a traffic packet to the MAC control unit 136.

  When the route table indicates a logical port, that is, an LA group, as the output destination port, the layer 2 control unit 131 selects the output destination physical port 105 from the LA group. The processing after selecting the output destination physical port 105 is the same as when the path table indicates the physical port 105 as the output destination port.

  In the following, processing for selecting the port 105 for transmitting a traffic packet from the LA group will be described. In the LA group, the layer 2 control unit 131 (specifically, the port selection unit 315) uses a plurality of ports (also referred to as LA use ports) that are used to transmit traffic packets and distribute traffic packets from ACT ports. , Based on the port status of the ACT port. Further, the layer 2 control unit 131 (specifically, the port selection unit 315) selects an output destination port of one received traffic packet from the LA use ports according to a predetermined load distribution algorithm.

  The layer 2 control unit 131 can use any load distribution algorithm, but can use, for example, a round robin algorithm, an address-based algorithm, or a port-based algorithm. In the round robin algorithm, for example, traffic packet transmission ports are selected in the order of the port numbers of the LA use ports.

  The address-based algorithm generates a hash value from part or all of the destination address of a traffic packet, and selects a port associated with the hash value from the LA use ports. As the destination address, for example, any one of a MAC address, an IP address, or a TCP port number is used. The port-based algorithm selects a port associated with a port that has received a packet from LA-used ports.

  In the example described below, the layer 2 control unit 131 reselects the LA use port, triggered by a change in the port state in the LPI control of any port 105 in the LA group. Thereby, an appropriate LA use port can be selected according to the current LA port state. When the LPI control unit 134 changes the port state of any one of the ports 105, the LPI control unit 134 notifies the layer 2 control unit 131 of the change. The layer 2 control unit 131 may or may not reselect the LA use port triggered by a change in the port state due to a failure.

  With reference to the flowchart of FIG. 7, the notification of the port state change by the LPI control unit 134 will be described. The port state change notification unit 345 of the LPI control unit 134 performs this process. When the state of any of the ports 105 is changed in the LPI control protocol processing (S101: YES), the port state change notification unit 345 changes the value of the port state column 352 of the corresponding entry in the port state management table 135. (S102).

  The port state change notification unit 345 further notifies the layer 2 control unit 131 of the port state change (S103). Specifically, the port state change notifying unit 345 transmits the port number and the value indicating the new port state of the port whose port state has been changed to the layer 2 control unit 131. The layer 2 control unit 131 searches the LA state management table 137 for the received port number, and updates the value of the port state column 373 of the corresponding entry to a value indicating a new state.

  When the layer 2 control unit 131 receives the notification of the port state change from the LPI control unit 134, the layer 2 control unit 131 reselects the LA use port in the LA group including the port. FIG. 8 is a flowchart illustrating an example of LA port reselection processing by the layer 2 control unit 131. The port selection unit 315 of the layer 2 control unit 131 performs this process.

  In the flowchart of FIG. 8, the port selection unit 315 monitors the port status change notification from the LPI control unit 134 (S201). When the port state change notification is received (S201: YES), the port selection unit 315 reflects the notified port state change in the LA state management table 137. Specifically, the notification indicates the port number and the new port state, and the port selection unit 315 changes the value of the port state column 373 in the entry of the port number.

  Next, the port selection unit 315 reselects the LA use port from the LA group including the notified port. First, the port selection unit 315 selects an ACT port as an LA use port candidate in the ports belonging to the LA group (S203). Specifically, the port selection unit 315 is the LA number corresponding to the port number notified of the value of the LA number column 371 from the LA state management table 137, and the value of the LA state column 374 is “ACT”. Select the port number for an entry.

  Next, the port selection unit 315 specifies the link aggregation mode of the LA group (S204). Specifically, the port selection unit 315 acquires the value of the LA group entry from the LA mode column 382 of the LA setting management table 138.

  When the link aggregation mode of the LA group is not the LACP mode (S204: NO), that is, the static mode, the port selection unit 315 identifies the load distribution port mode of the LA group (S205, S206). Specifically, the port selection unit 315 acquires the value of the LA group entry from the load balancing port mode column 384 of the LA setting management table 138.

  When the load balancing port mode is “default” and neither “power saving priority” nor “performance priority” (S205: NO, S206: NO), the port selection unit 315 determines the LA use port selected in step S203. Candidates are determined as LA use ports. That is, a port whose LA state is “ACT” is an LA use port, and an output destination port that transmits a traffic packet is selected according to the load distribution algorithm.

  When the selected LA use port includes a port state other than the active state, the port selection unit 315 may instruct the LPI control unit 134 to change the port state to the active state. In this case, the layer 2 control unit 131 does not need to instruct the port in the wake state to change to the active state.

  In step S204, when the LA group link aggregation mode is the LACP mode (S204: YES), the port selection unit 315 narrows down the LA use port candidates (S207).

  Specifically, the port selection unit 315 newly selects a port for which a LACP link has been established among the LA use port candidates selected in step S203 (the port whose value in the LA state column 374 is “ACT”). To select a candidate for the LA use port. The port selection unit 315 refers to the LA state column 374 and the LACP state column 375 of the LA state management table 137, and acquires the port number of the entry whose LA state is “ACT” and whose LACP state is “established”. . After step S207, the layer 2 control unit 131 proceeds to step S205. This avoids failure link selection.

  In step S205, when the load balancing port mode designated for the LA group is “emphasis on power saving” (S205: YES), the port selection unit 315 changes the port status from the LA use port candidate to those port states. Based on this, the LA use port is selected (S208). In the “power saving emphasis” mode, the ports in the active state and the wake state are selected as LA use ports with priority over other ports.

  Specifically, the port selection unit 315 selects a port whose port state is active or wake among the LA use port candidates selected in step S203 or step S207. As a result, it is possible to reduce power consumption caused by changing the port state to the active state. For example, the port selection unit 315 selects all ports in the active state and the wake state from the LA use port candidates as LA use ports. Thereby, the effect of load distribution can be enhanced.

  The port selection unit 315 refers to the LA state management table 137 and selects the port number of the entry whose value in the port state column 373 is “ACTIVE” or “WAKE” in the LA use port candidate entry. These are LA use ports to which transmitted traffic packets are distributed.

  The port selection unit 315 may select only a part of the ports in the active state and the wake state, for example, only the ports in the active state, from among the LA use port candidates.

  In step S206, when the load distribution port mode designated for the LA group is “performance-oriented” (S205: NO, S206: YES), the port selection unit 315 selects the LA use port candidate from Based on the actual traffic volume of the LA group, the number of ports satisfying appropriate communication performance is selected as the LA use port (S209).

  FIG. 9 is a flowchart showing details of the performance-oriented port selection step (S209) in the flowchart of FIG. The port selection unit 315 selects all ports in which the port state is active or wake among the LA use port candidates selected in step S203 or step S207 (S301). Specifically, the port selection unit 315 refers to the LA state management table 137 and selects the port number of the entry whose values in the port state column 373 are “ACTIVE” and “WAKE” from the LA use port candidate entries. .

  The port selection unit 315 compares the total bandwidth of the new LA use port candidate selected in step S301 with the transmission traffic amount of the LA group (S302). The port selection unit 315 holds port bandwidth management information indicating the bandwidth of each port 105. For example, the user transmits these values using the management computer, and the layer 2 control program 121 receives these values and passes them to the layer 2 control unit 131. The sum of the bandwidths of the LA use port candidates is the total bandwidth of the LA use port candidates.

  Further, the port selection unit 315 holds traffic information for managing the actual traffic volume of each LA group. For example, the port selection unit 315 monitors the output destination port (physical port or logical port) of the traffic packet determined by referring to the route table, and determines the relationship between each output destination port and the transmission traffic amount as traffic amount management information. Record in. The calculation method of the traffic volume to be managed is arbitrary depending on the design. For example, in the traffic management information, the moving average value of the transmission traffic amount within a predetermined period length indicates the traffic amount of each output destination port.

  When the total bandwidth of the LA use port candidate is larger than the recorded traffic volume of the LA group (S302: YES), the port selection unit 315 ends the flow. That is, the port selection unit 315 determines the current LA use port candidate as the LA use port.

  When the total bandwidth of the LA use port candidate is equal to or less than the recorded traffic volume of the LA group (S302: NO), the port selection unit 315 determines the port in the LA use port candidate not selected in step S301. It is determined whether there is a port whose state is quiet (S303). The port selection unit 315 refers to the LA state management table 137 and searches for an entry in which the value of the port state column 373 is “QUIET” in the LA use port candidate entries not selected in step S301.

  When there is a quiet port (S303: YES), the port selection unit 315 transmits a control message to the LPI control unit 134 so as to change the port state of the port to the active state (S306). When there are a plurality of quiet ports, the port selector 315 instructs the LPI controller 134 to change the port state of some or all of the ports to the active state.

  After step S306, the port selection unit 315 returns to step S301, adds the port changed to the active state in step S306, and redefines the LA use port candidate. Thereafter, the layer 2 control unit 131 proceeds to step S302.

  When the determination result in step S303 is negative (S303: NO), the port selection unit 315 determines whether there is a port whose port state is in the refresh state in the LA use port candidates not selected in step S301. (S304). The port selection unit 315 refers to the LA state management table 137 and searches for an entry in which the value of the port state column 373 is “REFRESH” in the LA use port candidate entries not selected in step S301.

  When there is a refreshed port (S304: YES), the port selection unit 315 transmits a control message to the LPI control unit 134 so as to change the port state of the port to the active state (S306). When there are a plurality of ports in the refresh state, the port selection unit 315 instructs the LPI control unit 134 to change the port state of some or all of the ports to the active state.

  When the determination result in step S304 is negative (S304: NO), the port selection unit 315 determines whether there is a port whose port state is in the sleep state among the LA use port candidates not selected in step S301. (S305). The port selection unit 315 refers to the LA state management table 137 and searches for an entry in which the value of the port state column 373 is “SLEEP” in the LA use port candidate entries not selected in step S301.

  If there is a port in the sleep state (S305: YES), the port selection unit 315 transmits a control message to the LPI control unit 134 so as to change the port state of the port to the active state (S306). When there are a plurality of ports in the sleep state, the port selection unit 315 instructs the LPI control unit 134 to change the port state of some or all of the ports to the active state.

  If the determination result in step S305 is negative (S305: NO), the port selection unit 315 ends this flow.

  In the “performance-oriented” flow, an appropriate number of ports can be selected according to the traffic volume measurement result, and the power consumption reduction effect by the LPI control can be enhanced. In the “performance-oriented” flow, the ports in the active state and the wake state are selected as LA use ports with priority over the ports in other port states. In other port states, the priority of selection is quiet, refresh, and sleep.

  Thus, in the port state transition order, by selecting from the port states that are close to the active state (the transition time and the number of states are small until the active state), it is possible to reduce power consumption due to the change of the port state. In the above example, the priority of the active state and the wake state in selecting the LA use port is the same, but the active state may be given priority over the wake state. Further, the priority of the refresh state and the quiet state may be the same.

  FIG. 10 schematically shows another configuration example of the packet relay device 10A to which the present invention can be applied. FIG. 10 shows an example of a chassis type packet relay device. The user can increase or decrease the network interface unit (number of ports) in the chassis type packet relay device 10A. Hereinafter, differences from the box-type packet relay apparatus shown in FIG. 1 will be described.

  In FIG. 10, the chassis type packet relay device 10 </ b> A includes a plurality of network interface units 108. The network interface unit 108 includes a port distribution control unit 133 and a plurality of ports 105, respectively. In the configuration shown in FIG. 10, the identifier of each port is composed of, for example, the number of the network interface unit 108 and the port number. The configuration of the port allocation control unit 133 is the same as that of the port allocation control unit 133 shown in FIG.

  Chassis-type packet relay device 10A includes a plurality of packet processing units 1003. Unlike the packet processing unit 103 in FIG. 1, the packet processing unit 1003 includes a relay processing unit 139. The relay processing unit 139 performs data transfer between the network interface units 108. Each packet processing unit 1003 performs ring aggregation based on the port state in transmission of a traffic packet from the network interface unit 108 in charge. The configuration of the layer 2 control unit 131 is the same as that of the layer 2 control unit 131 shown in FIG.

  As mentioned above, although embodiment of this invention was described, this invention is not limited to said embodiment. A person skilled in the art can easily change, add, convert, and delete each element of the above-described embodiments within the scope of the present invention.

  In the above example, the LA use port is reselected by using the change of the port state as a trigger, but the communication apparatus may use a different trigger. For example, the communication device may reselect the LA use port every time a predetermined time elapses or reselect the LA use port in response to reception of a traffic packet.

  In the above example, the LA use port is reselected with a change in all the port states as a trigger. However, the LA use port may be reselected only for some ports or some state changes. For example, the port state change between the refresh state and the quiet state may be ignored.

  In the above example, the layer 2 control unit 131 manages all the states in the LPI control in the LA state management table 137, but may manage only some of those states. For example, the LA state management table 137 may not indicate the refresh state. Further, the present invention can be applied to a communication apparatus that manages only a normal power state and a power saving state.

  The above example provides three load sharing modes including a default mode, but two or less or more than three load sharing modes may be provided. In addition, the communication device may provide only one LA mode. In the above example, the LA group can include the SBY port, but in other examples, the SBY port may not be prepared. The present invention can be applied not only to packet relay apparatuses but also to edge communication apparatuses in network communication, and can be applied to power control of ports different from EEE.

10A, 10B packet relay device, 102 memory, 103 packet processing unit, 104 bus, 105A-105D port, 106A-106D network line, 108 network interface unit, 121 control program, 131 layer 2 control unit, 132 packet buffer, 133 port Distribution control unit, 134 LPI control unit, 135 port state management table, 136 MAC control unit, 137 LA state management table, 138 LA setting management table, 139 relay processing unit, 315 port selection unit, 345 port state change notification unit, 1003 Packet processor

Claims (10)

Multiple physical ports,
A plurality of processing circuits that process packets of the plurality of physical ports; a first control unit that controls power supply to each of the plurality of processing circuits based on traffic of the plurality of physical ports;
First information for managing states of the plurality of physical ports in power supply control of the plurality of processing circuits by the first control unit;
A second control unit,
The second controller is
Combine the plurality of physical ports into one group and use the group as one logical port;
With reference to the first information, based on the state of the plurality of physical ports, a plurality of used physical ports for distributing transmission packets are selected from the group,
A communication device that selects a physical port for transmitting the packet from the plurality of used physical ports in transmission of the packet via the logical port. The communication device according to claim 1,
In the selection of the plurality of used physical ports, the second control unit sets a physical port in a normal power state and a physical port in a transition state from a power saving state incapable of packet transmission to the normal power state. A communication device that selects in preference to a physical port. The communication device according to claim 2,
Each of the selected plurality of used physical ports is a communication device that is the physical port in the normal power state or the transition state. The communication device according to claim 2,
The second control unit is a communication device that selects all the physical ports in the normal power state and the transition state as the plurality of used physical ports. The communication device according to claim 2,
The second controller is
Based on the measurement result of the traffic volume of the logical port and the bandwidth information of the plurality of physical ports, so that the total bandwidth of the plurality of used physical ports covers the traffic volume of the measurement result, A communication device that selects the plurality of used physical ports. The communication device according to claim 5,
The second control unit selects the physical port in the power saving state in preference to the physical port in the transition state from the normal power state to the power saving state in the selection of the plurality of used physical ports. Communication device. The communication device according to any one of claims 1 to 6,
The first control unit notifies the second control unit of a change in the state of the plurality of physical ports in power consumption control of the plurality of processing circuits by the first control unit;
In response to the notification, the second control unit reselects a plurality of used physical ports from the group. The communication device according to any one of claims 1 to 6,
The group includes a plurality of active physical ports and a standby physical port that is a switching destination of the active physical port in which a failure has occurred,
The second control unit is a communication device that selects the plurality of used physical ports from the plurality of operating physical ports. A communication apparatus comprising: a plurality of physical ports; and a plurality of processing circuits that process packets of the plurality of physical ports, and controlling power consumption of the plurality of processing circuits based on traffic of the plurality of physical ports. A control method,
Combine the plurality of physical ports into one group and use the group as one logical port;
Based on the state of the plurality of physical ports in power consumption control of the plurality of processing circuits, from the group, select a plurality of physical ports to be used to distribute transmission packets
A method for controlling a communication apparatus, wherein, in transmitting a packet via the logical port, a physical port for transmitting the packet is selected from the plurality of used physical ports. A communication device control method according to claim 9, comprising:
In the selection of the plurality of physical ports used, the physical port in the normal power state and the physical port in the transition state from the power saving state incapable of packet transmission to the normal power state have priority over the physical ports in other states. A communication device control method is selected.

Images