JP4751213B2 - Link aggregation processing apparatus and processing method - Google Patents

Description

  The present invention relates to a link aggregation processing apparatus and processing method, and more particularly to a link aggregation processing apparatus and processing method for performing link aggregation processing in a multi-blade layer 2 switch or layer 3 switch.

  Link aggregation is a protocol that bundles a plurality of physical ports and treats them as if they were one physical port for the purpose of physical link redundancy and port bandwidth enhancement. Etc.).

  When a packet is output by link aggregation, the packet may be output from any of the bundled physical ports, but the same communication, for example, a source address (SA) and a destination address (DA: Destination Address) ) Are similar, the packet needs to be output from the same physical port as long as communication continues. This is to prevent the packet order from being switched from being output from a plurality of physical ports.

  When setting link aggregation across blades, the output destination of a packet is normally determined on the input side. However, multicast packets have a problem of depleting internal transmission path resources.

  A conventional example will be described below with reference to FIG.

  FIG. 9 is a diagram illustrating a configuration of a layer 2 switch device including a plurality of blades (line cards) 81 to 84, a control unit 85, and a switch unit (SW) 86.

  The blade 81 includes a network interface unit (NIF unit) 811, a packet processing unit 812, and a table 813. These are also provided in the blades 82 to 84, respectively, but only the blade 81 is shown in FIG. The network interface unit 811 includes physical ports # 1 to # 3, which are also provided in the blades 82 to 84, respectively.

  The blades 81 to 84 are assigned PCU (Packet Control Unit) # 1 to PCU # 4 as device numbers.

  In the following description, the present invention will be described on the assumption that a VLAN (Virtual Local Area Network) and a LAG (Link Aggregation Group) as shown in FIG. 10 are used. FIG. 10 shows the physical port # 3 of the device number PCU # 1 indicating the blade 81 as LAG # 1 to LAG # 3, the physical port # 1 of PCU # 2, the physical port # 1 of PCU # 3, and PCU # 2. Physical port # 2 and physical port # 2 of PCU # 3, physical port # 3 of PCU # 2, and physical port # 3 of PCU # 3 are set, and the contents are stored in table 813. .

  The packet processing unit 812 is connected to the control unit 85, the switch unit 86, the network interface unit 811 and the table 813 in the same blade, and refers to the stored contents of the table 813 according to the control of the control unit 85, and transmits the packet. The physical port of the network interface unit 811 to be transmitted is determined.

  In a general switch device, the bandwidth of the transmission path in the blade is almost equal to the total bandwidth of the physical port (in any case, it is not significantly larger), and the packet flow rate is transmitted in the blade by the multicast packet for LAG. It is necessary to limit the amount of packets to be transferred and the number of LAGs that can belong to a VLAN so that the bandwidth of the path is exceeded and packet drops occur or the bandwidth is not exceeded in advance.

  Hereinafter, a conventional technique for packet transfer will be described.

Method 1
In both unicast packets and multicast packets, for each LAG to be output, after determining which physical port in the LAG is to be transmitted on the input side, an identifier indicating the port on the output side (hereinafter, output destination identifier) is packetized. To the output destination blade where the physical port exists.

  In the multicast packet transfer by the technique 1, as shown in FIG. 11, the transfer packets for the number of LAGs (for the number of ports) to be flooded by the packet processing unit 812 of the blade 81 on the input side are duplicated, and the SW unit 86 is To the blades 82 and 83 on the output side. Since an output identifier is assigned to the transfer packet at this time, the packet processing units (not shown) of the blades 82 and 83 simply send out to the port indicated by the output identifier without performing a duplication process or the like.

  In the multicast packet transfer according to the above technique 1, when a plurality of LAGs belong to the same VLAN and a plurality of output destination physical ports of these LAGs exist in a specific blade, the number of packets increases, The bandwidth of the transmission path between the SW unit is compressed. As a technique for preventing this, there is the following technique 2.

Method 2
Unicast packet transfer is the same as in Method 1, but multicast packet transfer is combined into one packet when there are multiple transfer packets for the same blade, and multiple output destination identifiers are assigned in the packet. To the output destination blade.

  In the transfer of the multicast packet by the method 2, as shown in FIG. 12, the packet processing unit 812 of the blade 81 on the input side transfers it to the blades 82 and 83 on the output side.

  The packet transferred to the blade 82 and the packet transferred to the blade 83 have different output identifiers. Since the packet transferred to the blade 82 has two output identifiers, The packet processing unit 822 performs duplication processing and sends it to the port indicated by each output identifier.

That is, the number of packet duplications is the same as the number of output side blades, but the length of the packet to be transferred increases according to the number of output side ports.
In Method 2, although packet duplication is limited to the number of output destination blades, the packet length of one packet is increased, which leads to compression of the bandwidth of the transmission path of the blade through which the packet with the increased packet length passes.

  As a technique for preventing these, there is the following technique 3.

Method 3
The unicast packet is the same as in the method 1, but in the multicast packet, the physical port in the LAG is determined on the output side. In this case, since it is not known which physical port of which blade is output on the input side, it is transferred to all blades on the output side that have the possibility of output (physical ports included in LAG belong) To do.

The packet processing unit provided in the output side blade transmits the received packet and the transmission state of each port to the control unit 85 (see FIG. 9), and the control unit 85 determines the output state of each physical port of each transmission side blade. The physical port for transmitting the packet is determined so as to be uniform, and the information of the output destination port of the packet is rewritten.
JP 2005-20492 A

  Among the conventional technologies described above, the methods 1 and 2 have the following problems.

  When a plurality of LAGs belong to the same VLAN and a plurality of output destination physical ports of these LAGs exist in a specific blade, multicast packet increases the number of packets in method 1, and the connection between the blade on the input side and the SW unit The bandwidth of the transmission line is pressed.

  In Method 2, although packet duplication can be suppressed to the number of output destination blades, the packet length of one packet is increased, so the bandwidth of the transmission path (between the packet processing unit and the switch unit) in the blade is compressed.

  Further, method 3 has the following problems.

  Depending on the packet type (unicast or multicast), the location where the output destination port is determined changes, which complicates the processing. In addition, when moving the output destination port, it is realized by directly rewriting the information of the output destination port. However, in the multicast packet, the output destination port between the blades in the process of rewriting the information of the output destination port of each output side blade. Therefore, the information is temporarily not matched, so that packets are instantaneously transmitted (double transmission) from a plurality of physical ports, or packets are not transmitted from any physical port.

  The present invention has been made in view of the problems of the conventional techniques as described above, and an object of the present invention is to realize efficient processing in a multicast packet and improve the problems of existing methods. .

In the link aggregation processing apparatus of the present invention, a plurality of blades having one or more physical ports are connected via a switch unit, and the link aggregation is configured such that there are a plurality of physical ports of the blade on the output side. In the processing device,
For each blade,
A packet processing unit and a table are provided;
In the table,
A forwarding list table that exists for each VLAN and holds information about the distribution destination of the blade on the output side and multicast ID information for each destination address;
A LAG key table for storing LAG keys;
An output port map table for mapping the multicast ID and the output port;
A LAGch table that maps the physical port of the blade on the output side corresponding to the LAG key and the output port;
The packet processing unit of the blade on the input side extracts the VLAN and the destination address when a packet is input, and assigns the distribution destination information and the multicast ID stored in the forwarding list table corresponding to these, Furthermore, the LAG key stored in the LAG key table is attached to the transmission packet and transmitted to the switch unit,
In the switch unit, the transmission packet is duplicated and transmitted to the output blade according to the output blade shown in the distribution destination information,
The packet processing unit of the blade on the output side refers to the output port map table, obtains an output port corresponding to the multicast ID assigned to the transmission packet, and then refers to the LAGch table, Obtain the physical port of the blade on the output side that outputs from the LAG key and output port, and if the physical port belongs to the own blade, send the packet from that port and send it to the other blade If it belongs to the port, the packet is discarded.
It is characterized by that.

  In this case, the control unit includes a control unit that controls each packet processing unit so that the usage state of the physical port of the blade on the output side is averaged, and the control unit includes the physical port from which the transmission packet is output. When changing, the LAG key referred to by the transmission packet is obtained. Next, an unused entry common to all LAGs of all blades is determined in the LAGch table, and the index value of the entry is set as a new index value. The LAG that moves to the use entry sets the destination physical port, the non-moving LAG sets the currently used physical port, and then updates the value of the LAG key on the input side to the new index, and the packet It is good also as giving to.

  Further, the control unit may perform control based on the transmission content from the packet processing unit of the output side blade.

  The control unit may perform control based on a transmission packet from the packet processing unit of the blade on the input side to the switch unit.

  The control unit may perform control based on a transmission packet from the switch unit to the packet processing unit of the blade on the output side.

  In any of the above, the packet processing unit of the blade on the input side may determine the LAG key to be given to the transmission packet by a distributed algorithm.

The link aggregation processing method of the present invention is configured such that a plurality of blades having one or more physical ports are connected via a switch unit, and the physical port of the blade on the output side is plural.
For each blade,
A packet processing unit and a table are provided;
In the table,
A forwarding list table that exists for each VLAN and holds information about the distribution destination of the blade on the output side and multicast ID information for each destination address;
A LAG key table for storing LAG keys;
An output port map table for mapping the multicast ID and the output port;
In a link aggregation processing method performed by a link aggregation processing device provided with the LAG key and a LAGch table that maps the physical port of the blade on the output side corresponding to the output port,
When a packet is input, the packet processing unit of the blade on the input side extracts the VLAN and destination address, and assigns the distribution destination information and multicast ID stored in the forwarding list table corresponding to these, A step of assigning a LAG key stored in the LAG key table to a transmission packet and transmitting the packet to the switch unit;
The switch unit replicates the transmission packet according to the output side blade indicated in the distribution destination information and transmits it to the output side blade;
The packet processing unit of the blade on the output side refers to the output port map table, obtains an output port corresponding to the multicast ID assigned to the transmission packet, and then refers to the LAGch table, Obtain the physical port of the blade on the output side that outputs from the LAG key and output port, and if the physical port belongs to the own blade, send the packet from that port and send it to the other blade If the port belongs to, the packet has a step of discarding;
It is characterized by that.

  In this case, when changing the physical port from which the transmission packet is output when controlling through each packet processing unit so that the usage status of the physical port of the blade on the output side is averaged, The LAG key referred to by the transmission packet is obtained. Next, an unused entry common to all LAGs of all blades is determined in the LAGch table, and the LAG that moves to the unused entry with the index value of the entry as a new index value is obtained. The destination physical port is set, the non-moving LAG has a physical port that is currently used, and then the value of the LAG key on the input side is updated to a new index and added to the packet. It is good.

  The present invention configured as described above has the following effects.

The following improvements can be made in multicast packets.
(1) Band consumption of the transmission unit to the LAG can be improved, and packet discard does not occur.
(2) Even if the number of LAGs in the VLAN increases, the bandwidth used by the transmission unit does not increase.
(3) The number of LAGs in the VLAN is not limited.
(4) The output destination determination process concentrated on the input side can be distributed to each blade on the output side.
(5) The layer 2 switch device can be configured without significantly increasing the bandwidth of the transmission unit with respect to the total bandwidth of the physical ports.
(6) When the output destination port is moved, double transmission and non-transmission of packets can be prevented.

  Next, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram showing a configuration of an embodiment of a layer 2 switch device according to the present invention. In FIG. 1, blades 11 to 14, control unit 15, switch unit 16, blade 11 are network interface unit 111, packet processing unit 112, and table 113 are blades 81 to 84 and control unit shown in FIG. 85, the switch unit 86, and the blade 81 are the same as the network interface unit 811, the packet processing unit 812, and the table 813. In the present embodiment, the table 113 includes the tables shown in FIGS. 2 to 5 in addition to the VLAN configuration shown in FIG.

  The forwarding list table shown in FIG. 2 exists for each VLAN, and holds information on the output destination interface, presence / absence of LAG key notification, and multicast ID for each destination address (DA). The output destination interface includes the PCU number, physical port, and distribution destination information by the switch unit. Among these, the PCU number and physical port number may be indicated as LAG numbers. There is one entry for multicast in each table. Refer to the input blade.

  For example, if the destination address is A, the value of the output destination interface is 1 for the PCU, 3 for the physical port, 1 for the SW distribution destination, 1 for the multicast ID, and x for the LAG key notification. When the destination address is B, the value of the output destination interface is LAG # 1, the SW distribution destination is 2, the multicast ID is no value, the LAG key notification is ◯, that is, notification is entered.

  The LAG key table shown in FIG. 3 is a table that exists in a blade and maps a reference value that is a specific value (Hash value) and a LAG key. The LAG key is a common key used for determining an output destination port on the output blade by each LAG. The value in the LAG key table may vary between blades. Refer to the input blade. In the example shown in FIG. 3, LAG keys 0 to M (M and N are arbitrary numbers) are mapped corresponding to Hash values 0 to N.

  The output port map table shown in FIG. 4 is a table that maps a multicast ID (multicast ID for each VLAN) and an output port. Refer to the output blade. In the example illustrated in FIG. 4, 1: 3, LAG # 1, LAG # 2, and LAG # 3 are mapped as output ports for the multicast ID “0”. This output port indicates the VLAN shown in FIG. 10, and all the ports having a link relationship are described.

  The LAGch table shown in FIG. 5 is a table for mapping the LAG key and the output port for each LAG, and is referred to by the output side blade. In the example illustrated in FIG. 5, 2: 1, 2: 2, and 3: 3 are described for each of LAG # 1, LAG # 2, and LAG # 3 with respect to “0” of the LAG key.

  In the present embodiment, the following means are used.

  Means 1: The LAG output destination port is determined by the output blade regardless of the packet type.

  Specifically, the packet processing unit provided on the input side blade refers to the LAG key table shown in FIG. 3 to determine the output destination port on the output side blade by each LAG. The LAG key table reference value is obtained by being distributed, and the LAG key corresponding to the reference value is given to the transfer packet and transmitted to the output side.

  This operation is performed in order to use the output destination ports on the output side blades on an average (distributed) basis, and is not necessarily limited to this method. A method of selecting a LAG key to be used at random or a method of selecting in order may be considered.

  The packet processing unit provided in the output blade refers to the LAGch table, acquires the output destination port using the passed LAG key as an index, and determines whether to transmit or discard the packet.

  Note that the number of entries in the LAGch table is set to be larger than the number of entries in the LAG key table in consideration of the movement of the output destination port described later.

  Means 2: Packet distribution destination information is added to the packet to be passed to the switch unit.

  In the unicast packet, the packet processing unit provided in the input side blade specifies the blade where the output destination port exists, and in the multicast packet, as a transmission to each blade having a physical port belonging to the flood target LAG Only one packet is transferred to the switch unit, and the switch unit duplicates the packet for the number of blades and transmits it to each blade.

  Means 3: The contents of LAGch of the same LAG are made to coincide with each other by synchronizing with all output blades.

  Specifically, it is realized by assigning only the index of the entry whose information matches in the LAGch table of all blades as the LAG key.

  That is, when moving the output destination port, after setting a new output destination port in a new entry (new index) in the LAGch table, the value of the LAG key in the LAG key table is updated to the new index value.

  FIG. 6 is a diagram schematically showing packet transfer performed in this embodiment. The packet processing unit 112 of the input-side blade 11 assigns switch distribution destination information and a LAG key to the input packet and sends it to the switch unit 16. The switch unit 16 duplicates as many as the number of output side blades indicated in the switch distribution destination information, and transmits to the blades 12 and 13 that perform output. The packet processing units 122 and 132 of the output side blades 12 and 13 refer to the LAGch table, acquire the output destination port using the passed LAG key as an index, and determine whether to transmit or discard the packet.

  Next, the packet transfer operation in this embodiment will be described with reference to FIG. FIG. 7 is a diagram mainly showing the operation of the packet processing unit in each blade.

  The packet processing unit of the input side blade 11 extracts the VLAN and destination address (DA) of the received packet of the input packet, and determines the output destination interface with reference to the forwarding list table shown in FIG. At this time, if the corresponding DA does not exist in the forwarding list table, it is regarded as a multicast packet, the multicast entry in the forwarding list table is referred to, and SW distribution destination information and multicast ID are given to the packet (operation 1121).

  Next, a specific value is generated using a specific distributed algorithm such as a hash, and a LAG key (2 in this embodiment) is obtained by referring to the LAG key table shown in FIG. 3 (operation 1122). The packet is attached to the packet (operation 1123) and sent to the switch unit 16.

  The SW unit 16 performs duplication based on the number of output side blades indicated in the distribution destination information given to the packet, and transfers them to the output side blades 12 and 13.

  The packet processing unit of the blades 12 and 13 on the output side refers to the output port map table, and outputs the output port corresponding to the multicast ID assigned to the transfer packet (output target LAG, LAG # 1 in this embodiment). ~ LAG # 3). Subsequently, referring to the LAGch table, a physical port that performs output from the LAG key and the output target LAG is acquired (operations 1221 and 1321). If the acquired output physical port is a port belonging to the own blade, the packet is transmitted from that port, and if it is a port belonging to another blade, the packet is discarded.

  In the case shown in FIG. 7, the output-side blade 12 transmits packets from the physical ports (2: 1) and (2: 2) belonging to LAG # 1 and LAG # 2, respectively, and belongs to LAG # 3 ( The transmission packet from 3: 3) is discarded.

  Further, the output blade 13 transmits a packet from the physical port (3: 3) belonging to LAG # 3, and the packets from (2: 1) and (2: 2) belonging to LAG # 1 and LAG # 2 respectively. The transmitted packet is discarded.

  The point of the above operation is that the packet transferred from the input blade to the output blade is only transferred one packet with a LAG key assigned to one blade regardless of the number of LAGs to be flooded. The decision is made on the output side.

  As a result, the length of information added to the transfer packet can be fixed without depending on the number of LAGs and the number of output blades in the VLAN, and efficient packet transfer is possible.

  The operation procedure of the unicast packet is the same as the operation procedure described above, except that the part where the SW distribution destination designates a specific blade is different on the input side, and other processes can be made common.

  Next, an operation when rewriting the information of the output destination port of the packet so that the output state of each physical port of each transmitting blade is uniform will be described with reference to FIG.

  The operations 1121 to 1123 shown in FIG. 7 and the duplication and transmission operations in the switch unit 16 are performed in the same manner.

  The packet processing units of the blades 12 and 13 on the output side transmit the output state of the physical port of the own blade to the control unit 15 shown in FIG. 1, and the control unit 15 outputs the output of each physical port of each blade according to this transmission content. The control is performed so that is uniformized. In the following description, (2: 1), (2: 2), and (3: 3) are set as LAG # 1 to LAG # 3 in the packet processing units of the blades 12 and 13 on the output side. It is assumed that the physical port that is reported and output is changed from (2: 1) to (3: 1).

  First, the LAG key of the LAG key table referred to by the packet to which the output physical port moves is obtained (2 in this case).

  Next, in the LAGch table, an unused entry common to all LAGs of all blades is determined (N in this case), and the LAG that moves to the unused entry with the index value of the entry as the new index value is The physical port is set, and the non-moving LAG sets the currently used physical port (operations 1222 and 1322).

  Next, the value (LAG key) in the LAG key table on the input side is updated to the new index (operation 1124), added to the packet (operation 1125), and sent to the switch unit 16.

  The above control is performed by the control unit 15 via the packet processing unit. The control unit 15 has been described as performing control based on the transmission contents from the packet processing unit of the output side blade. However, the control unit 15 includes a table similar to each blade, and the packet sent from the input side blade to the SW unit 16. Alternatively, the SW unit 16 receives a packet to be sent to the output side packet, and further refers to the processing status in the packet processing unit of the input side blade to grasp the usage status of the physical port of the output side blade and It is good also as what performs control.

  The point of the above operation is that the LAG key on the input side is updated after the preparation is completed without changing the LAG key until the information in the LAGch table of the output blade is completed.

  Thereby, it is possible to prevent the same packet from being transmitted twice or not transmitted when the output destination is moved.

It is a block diagram which shows the structure of one Example of the layer 2 switch apparatus by this invention. It is a figure which shows the structure of a forwarding list table. It is a figure which shows the structure of a LAG key table. It is a figure which shows the structure of an output port map table. It is a figure which shows the structure of a LAGch table. It is a figure which shows roughly the packet transmission performed in the Example of this invention. It is a figure which mainly shows operation | movement of the packet process part in each blade. It is a figure which mainly shows operation | movement of the packet process part in each blade. It is a block diagram which shows the structure of a prior art example. It is a figure which shows the structure of VLAN and LAG. It is a figure which shows schematically the packet transfer performed in a prior art example. It is a figure which shows schematically the packet transfer performed in a prior art example.

Explanation of symbols

11 to 14 Blade 15 Control unit 16 Switch unit 111 NIF unit 112 Packet processing unit 113 Table

Claims (8)

In a link aggregation processing device configured such that a plurality of blades having one or more physical ports are connected via a switch unit, and the plurality of physical ports of the blade on the output side are configured.
For each blade,
A packet processing unit and a table are provided;
In the table,
A forwarding list table that exists for each VLAN and holds information about the distribution destination of the blade on the output side and multicast ID information for each destination address;
A LAG key table for storing LAG keys;
An output port map table for mapping the multicast ID and the output port;
A LAGch table that maps the physical port of the blade on the output side corresponding to the LAG key and the output port;
The packet processing unit of the blade on the input side extracts the VLAN and the destination address when a packet is input, and assigns the distribution destination information and the multicast ID stored in the forwarding list table corresponding to these, Furthermore, the LAG key stored in the LAG key table is attached to the transmission packet and transmitted to the switch unit,
In the switch unit, the transmission packet is duplicated and transmitted to the output blade according to the output blade shown in the distribution destination information,
The packet processing unit of the blade on the output side refers to the output port map table, obtains an output port corresponding to the multicast ID assigned to the transmission packet, and then refers to the LAGch table, Obtain the physical port of the blade on the output side that outputs from the LAG key and output port, and if the physical port belongs to the own blade, send the packet from that port and send it to the other blade If it belongs to the port, the packet is discarded.
A link aggregation processing device characterized by that. The link aggregation processing device according to claim 1,
A control unit that controls each packet processing unit so that the usage status of the physical port of the blade on the output side is averaged, and the control unit changes the physical port from which the transmission packet is output The LAG key referred to by the transmission packet is obtained, and then an unused entry common to all LAGs of all blades is determined in the LAGch table, and the index value of the entry is used as a new index value. The moving LAG sets the destination physical port, the non-moving LAG sets the currently used physical port, and then updates the value of the LAG key on the input side to the new index and adds it to the packet A link aggregation processing device characterized by that. In the link aggregation processing device according to claim 2,
The link aggregation processing device, wherein the control unit performs control based on transmission contents from the packet processing unit of the blade on the output side. In the link aggregation processing device according to claim 2,
The link aggregation processing device, wherein the control unit performs control based on a transmission packet from the packet processing unit of the blade on the input side to the switch unit. In the link aggregation processing device according to claim 2,
The link aggregation processing device, wherein the control unit performs control based on a transmission packet from the switch unit to the packet processing unit of the output blade. In the link aggregation processing device according to any one of claims 1 to 5,
The link aggregation processing device, wherein the packet processing unit of the blade on the input side determines a LAG key to be given to the transmission packet by a distributed algorithm. A plurality of blades having one or more physical ports are connected via a switch unit, and the blades on the output side are configured to have a plurality of physical ports,
For each blade,
A packet processing unit and a table are provided;
In the table,
A forwarding list table that exists for each VLAN and holds information about the distribution destination of the blade on the output side and multicast ID information for each destination address;
A LAG key table for storing LAG keys;
An output port map table for mapping the multicast ID and the output port;
In a link aggregation processing method performed by a link aggregation processing device provided with the LAG key and a LAGch table that maps the physical port of the blade on the output side corresponding to the output port,
When a packet is input, the packet processing unit of the blade on the input side extracts the VLAN and destination address, and assigns the distribution destination information and multicast ID stored in the forwarding list table corresponding to these, A step of assigning a LAG key stored in the LAG key table to a transmission packet and transmitting the packet to the switch unit;
The switch unit replicates the transmission packet according to the output side blade indicated in the distribution destination information and transmits it to the output side blade;
The packet processing unit of the blade on the output side refers to the output port map table, obtains an output port corresponding to the multicast ID assigned to the transmission packet, and then refers to the LAGch table, Obtain the physical port of the blade on the output side that outputs from the LAG key and output port, and if the physical port belongs to the own blade, send the packet from that port and send it to the other blade If the port belongs to, the packet has a step of discarding;
A link aggregation processing method characterized by the above. The link aggregation processing method according to claim 7,
When changing the physical port from which the transmission packet is output when controlling via each packet processing unit so that the usage status of the physical port of the blade on the output side is averaged, the transmission packet is A LAG key to be referred to is obtained, and then an unused entry common to all LAGs of all blades is determined in the LAGch table, and the LAG moving to an unused entry with the index value of the entry as a new index value is The non-moving LAG sets the physical port that is currently used, and then updates the value of the LAG key on the input side to a new index and assigns it to the packet. Link aggregation processing method.

Images