JPH10341251A - Router and frame transfer method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a virtual identifier used for switching a data link layer by using information described in a prescribed area for an identifier to refer to a cross-reference stored in a storage means, providing address information of a node of a next stage obtained without network layer processing to a frame and transferring the resulting frame. SOLUTION: In the case of cut-through transfer, a MAC discrimination section 241 gives a received MAC frame to a flow processing section 251, and in the case of conventional network layer transfer, the section 241 gives the received MAC frame to a datagram processing section 201. The flow processing section 251 retrieves a next stage information table 261 based on the cut-through MAC address information and a flow identifier in a MAC frame to obtain information such as a network connection interface to which a frame is to be sent and cut-through identifier of a next stage or the like and to rewrite header information of a MAC frame as required.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a router capable of connecting communication networks such as Ethernet and transferring frames, and a frame transfer method.

[0002]

2. Description of the Related Art A router device has a different LAN (Local Area N).
etwork) to connect between one LAN
To transfer the datagram to the other LAN.
In a datagram, in addition to communication information to be transferred, a network layer address of a transmission source and a final destination thereof (for example, an IP address in an IP (Internet Protocol)) is described. The datagram output network interface and the destination node (router device or host) are determined using the datagram.

In a conventional router, a data link layer (for example, a MAC (Media Access Control)) is used instead of a LAN.
When a data link layer frame (MAC frame) defined in the layer is received, the frame is subjected to data link layer processing, and then converted into a packet (datagram) handled in the network layer. Then, based on the network layer address of the final destination described in the datagram, the output interface of the datagram and the network layer address of the next node to be transferred are determined, and the datagram is passed to the corresponding output interface. . In the output interface, after determining the data link layer address from the network layer address of the next node,
Construct data link layer frame and output to LAN.

[0004] In a router device for transferring communication information (data link frame) according to the above procedure, a received data link layer frame is converted into a network layer packet, and then the packet such as a final destination address and a source address is converted. , Based on the routing table held by the router, determine the node (next stage router or destination host) to which the packet should be sent next, the output interface, and convert it to a data link layer frame again. The operation of outputting to the next stage node is performed.

[0005] At this time, in the conventional router, excessive processing such as conversion of a received data link layer frame into a network layer packet, searching of a routing table, and conversion of a packet into a data link frame causes a conventional router to receive a packet. The processing time required for transferring the data link layer frame and the processing load inside the router are enormous.

[0006]

In order to reduce the load of the router processing, a method of determining a frame transfer destination at a data link layer level without referring to information at a network layer level described in a packet is known. It has been proposed (this transfer method may be hereinafter referred to as cut-through).

One is intermediate header information described in a data link layer frame (for example, LLC (Logical
Link Control) header, and the like, by adding a virtual identifier assigned in correspondence with the final destination information and the transmission source information, thereby enabling switching at the data link layer frame level.

[0008] That is, the router device that has received the frame inspects the LLC header to perform normal network layer datagram processing on the frame and then transfer the frame, or cut-through the data link layer frame as it is. Decide what to transfer. If cut-through transfer is to be performed, data link layer switching transfer is performed based on the virtual identifier added to the intermediate header information.

However, in this method, since a few bytes of information are newly added to the header information, the length of a portion in the frame in which the high-order datagram can be put becomes short. This means that the maximum packet length that can be used for switching in the data link layer frame is different from the maximum packet length that can be used for normal datagram transfer.
Then, when a datagram is sent from a source host or the like that does not support the above method, it is necessary to readjust the packet length at the first router. Even in a node that supports the above method, it is necessary to perform processing such as changing the packet length when switching from normal transfer to data link layer switching transfer in the network layer.

On the other hand, a method is also conceivable in which a virtual identifier required for data link layer switching is assigned using a local address field defined by a data link layer address used in a LAN.

That is, a dedicated virtual identifier is assigned to each of the packet flows to be subjected to cut-through transfer, and this virtual identifier is written in the destination MAC address field or the source MAC address field of the frame. Then, the router device that has received the frame whose MAC address is local performs data link layer switching transfer based on the virtual identifier described in the MAC address field.

When the virtual identifier is described in the destination MAC address field and used, it is necessary to store a new MAC address so that each receiving node (router or host) can receive the local destination MAC address. become. Therefore, if you want to perform data link layer switching for each of many packet flows,
Each node must store a number of virtual identifiers as MAC addresses to receive.

[0013] In this case, when an Ethernet switch is present between nodes such as Ethernet, the Ethernet switch can relay a frame in which a local destination MAC address is written to an appropriate node. A large number of MAC addresses must be stored.

When the virtual identifier is described in the field of the transmission source MAC address and used, the bridge device with the learning function forming the LAN learns many transmission source MAC addresses which do not need to be stored. This consumes a large amount of the MAC address storage area in the bridge device.

The present invention utilizes the switching at the data link layer frame level to improve the transfer efficiency of the router device, and it is also necessary to readjust or change the packet length by using a node, a bridge, an Ethernet switch or the like. The purpose is to provide a virtual identifier that can be used for data link layer switching so that it is not necessary to store many MAC addresses.

[0016]

A router device according to the present invention uses a frame to be transferred at a data link layer level, which is defined as a region for writing upper layer protocol identification information. Receiving means for receiving a frame in which information for identifying the packet flow to which the frame belongs in the data link layer is written; an identifier capable of identifying the packet flow; and a next stage to which the frame belonging to this packet flow should be transferred Storage means for storing a correspondence relationship with the address information of the node, and the correspondence relationship stored in the storage means as an identifier at least the information written in the area of the frame received from the preceding node by the receiving means. By referencing, the address information of the next-stage node obtained without performing the network layer processing is stored in the frame. Imparting to, characterized in that by comprising a transfer means of the data link layer level transfer.

Here, the receiving means includes means for receiving another frame to be transferred at a network layer level from a node belonging to a certain logical network, and the router device assembles the another frame into a packet, The packet may be further provided with a network-layer-level transfer means for subjecting the packet to network layer processing, forming a frame again and transferring the frame to a node belonging to another logical network.

Further, information indicating whether or not the address information written in the address area of the frame received by the receiving means includes the data link layer level dedicated to transfer is included in the frame. For example, from a plurality of bits at a predetermined position in the address area),
The information processing apparatus may further include means for transferring the frame to the data link layer level transfer means or the network layer level transfer means based on the detection result.

The storage means may include not only the identifier capable of identifying the flow to which the packet belongs, and information for specifying the next-stage node written in the destination address area, but also that the next-stage node may identify this flow. May be stored in association with a new identifier. In this case, the data link layer level transfer means writes information based on the new identifier in an area defined as an area for writing upper layer protocol identification information of the frame. The information written in the destination address area of the frame may be an address including a node identifier assigned to a next-stage node exclusively for data link layer level transfer. An address fixedly assigned to the node at the next stage for the transfer at the network layer level is entered, and information based on a part of the new identifier (the router device dedicated to the data link layer level transfer) is written in the source address area. Alternatively, an address including a node identifier assigned to the next-stage node) can also be entered.

On the other hand, the storage means can identify not only the identifier capable of identifying the flow to which the packet belongs, the information specifying the next-stage node written in the destination address area, but also the upper layer protocol of the packet. A correspondence relationship with the protocol identification information may be stored. In this case, the data link layer level transfer means writes the protocol identification information in an area defined as an area for writing the upper layer protocol identification information of the frame. In the destination address area, an address fixedly assigned to the next node for transfer at the network layer level can be entered.

The present invention (using a destination address area) uses a frame to be transferred at the data link layer level, which is defined as an area for writing upper layer protocol identification information. Receiving means for receiving a frame in which information for identifying the packet flow to which the frame belongs at the data link layer, an identifier capable of identifying the packet flow, and the next Storage means for storing a correspondence relationship with address information of a stage node; and a correspondence relationship stored in the storage unit as an identifier at least information written in the area of a frame received from the preceding node by the reception means. The address information of the next node obtained without performing the network layer processing is referred to Characterized by comprising a transfer means of the data link layer level transfer the frame to impart a destination address region of at least the frame.

In this case, means for storing an address including a node identifier allocated to the router device exclusively for the data link layer level transfer, and the receiving means writes the stored address in a destination address area. Means for controlling the received frame to be received.

Further, the data link switch which performs frame routing using the destination address as a key can perform frame routing using an address including a node identifier assigned to the router device exclusively for the data link layer level transfer as a key. Means for transmitting a frame in which the dedicated address is assigned to the transmission source address area.

The present invention (using a source address area) uses a frame to be transferred at the data link layer level, which area is defined as an area for writing upper layer protocol identification information. Receiving means for receiving a frame in which information for identifying the packet flow to which the frame belongs in the data link layer, an identifier capable of identifying the packet flow, and a frame belonging to this packet flow should be transferred. It is defined as storage means for storing a correspondence relationship with the address information of the next-stage node, and a source address area of a frame received from the previous-stage node by the receiving means and an area for writing the upper-layer protocol identification information. Refer to the correspondence stored in the storage means as information written in the area as an identifier A data link layer level transfer means for transferring the frame by adding the address information of the next node obtained without performing the network layer processing to a destination address area of the frame. And

At this time, the portion corresponding to the source address area in the identifier may include a node identifier assigned to the router device or the preceding node exclusively for data link layer level transfer. .

Here, the storage means stores not only the identifier capable of identifying the flow to which the packet belongs but also the address of the next-stage node, and that the next-stage node identifies this packet flow in the data link layer. A new identifier to be enabled (a node identifier assigned to the next-stage node or the router device exclusively for data-link-layer-level transfer can be stored) is stored, and the data-link-layer-level transfer means stores The new identifier may be written in an area defined as a source address area of a frame and an area for writing upper layer protocol identification information.

The invention relating to each apparatus of the present invention is also understood as the invention relating to the method. For example, the frame transfer method according to the present invention includes an identifier capable of identifying a flow to which a packet received from a certain logical network belongs at a data link layer, and another data link layer frame of a packet belonging to the flow to be transferred. If the correspondence relationship with the address information of the next node belonging to the logical network is stored, and the information written in the address area of the received frame indicates that the frame should be transferred at the data link layer level For example, by referring to the stored correspondence as an identifier, the information written in an area defined as an area where at least upper layer protocol identification information of the frame is written, without performing network layer processing The obtained address information of the next stage node is added to the frame and transferred. .
Further, if the frame is not to be transferred at the data link layer level, the frame is assembled into a packet,
The packet may be subjected to network layer processing, framed again, and transferred to another logical network.

The present invention is also understood as an invention relating to a storage medium storing a program for causing a computer to execute each means of the apparatus and each step of the method described above.

According to the present invention, a virtual identifier usable for data link layer switching can be carried using an address area and an area for upper layer protocol identification information. Is no longer necessary. Also, in the node or bridge, the Ethernet switch, etc., among the virtual identifiers required for data link layer switching, it is sufficient to store the information described in the address area, so that the MAC address that must be stored can be reduced. .

[0030]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows an example of a MAC address format and a data link layer frame (hereinafter abbreviated as MAC frame) format. The address format 101 (FIG. 1 (a)) is a normal MAC address format defined by the IEEE 802 committee, and includes a 1-bit group identifier (I / G bits), a 23-bit vendor identifier, and a 24-bit identifier.
Are classified into node identifiers. The first bit of the 23-bit vendor identifier is called U / L bit especially in the case of Ethernet, and is set to 0 in case of global use.

Address format 111 (FIG. 1 (b))
Is a MAC address format newly used to operate the present embodiment based on the address format 101. As the vendor identifier 112, a value that has been agreed in advance by the entire network as a value dedicated to cut-through is entered. With this vendor identifier 112, the received MAC
Each node can determine whether the frame is to perform cut-through transfer or normal datagram transfer.

The node identifier 113 is for identifying a node in each network when performing cut-through transfer. Even if a value is assigned so as to be unique in all LANs (entire network), A unique value may be assigned within each LAN. For the purpose of use, it is sufficient for each LAN as in the latter case. That is, each node:
Both a normal node identifier and a node identifier for cut-through will be assigned. In addition, one LAN
If all of the network interfaces are the same vendor (identifier), the node identifier for cut-through may have the same value as the normal node identifier.

MAC frame format (FIG. 1 (c))
Shows the frame format used in 802.3 as an example of the MAC frame format defined by the IEEE802 committee. MAC frame 121 is sent to destination MA
After the C address, the source MAC address, and the 2-octet frame length field, the LLC header identifies the upper layer protocol described below. For example, in the case of an IP datagram, encapsulation of a total of 8 octets called LLC / SNAP is used.

MAC frame format (FIG. 1 (d))
Is a DIX format MAC frame defined as Ethernet Version 2 on Ethernet.
IP uses the MAC frame format shown in Fig. 1 (d) as the default. MAC frame 122
After the MAC address and the source MAC address, the upper layer protocol is identified by a 2-octet Type field.

In the case of Ethernet, the MAC frame 121 and
In the MAC frame 122, the maximum datagram length (hereinafter referred to as the MTU size) that can be used by the upper layer protocol is different.
1492 bytes for C frame 121, MAC frame 122
Is 1500 bytes.

The MAC frame format 131 (FIG. 1 (e)
) Is M which is newly defined in order to operate this embodiment.
It is an example of an AC frame format. The MAC frame format used in this embodiment is such that a flow identifier 13 is added after a destination MAC address 132 and a source MAC address 133.
This format inserts 4 according to the address format 111 on the destination MAC address side, and inserts the format according to the address format 111 on the source MAC address side. Here, the destination MAC address or the source MAC according to the address format 111
The combination of the address and the Flow identifier 134 is called a cut-through identifier. The cut-through identifier is an example of a virtual identifier required for data link layer switching.

The Flow identifier is used in combination with an upstream or downstream cut-through node identifier so that an upstream node and a downstream node can uniquely identify a packet flow to be subjected to cut-through transfer in the data link layer. is there. In other words, the node that has received the frame can know from the data link layer which packet flow the frame belongs to by looking at the Flow identifier included in the frame and, if necessary, the node identifier for cut-through, and the network layer. The frame can be transferred without analyzing the flow to which the packet belongs by the processing.

The general flow identification information used to identify a packet flow at the network layer is unique across the entire network, whereas Fl
The ow identifier is sufficient if the upstream node and the downstream node sharing the ow identifier can identify a packet flow using the ow identifier.

For this reason, prior to actually transferring a frame by cut-through, a cut-through identifier including a Flow identifier, identification information at a network layer level of a packet flow represented by the cut-through identifier, and a flow identifier are defined between an upstream node and a downstream node. Exchange messages that associate The value of the cut-through identifier may be determined by the upstream node and notified to the downstream node, may be determined by the downstream node and notified to the upstream node, or may be determined by the downstream node in response to the request of the upstream node. You may reply to the node.

The cut-through identifier is defined so that a packet flow to be subjected to cut-through transfer can be uniquely specified at least in each network. When one cut-through node identifier is assigned to one node, it is possible to form a cut-through transfer virtual path exclusively used by each packet flow of the number of Flow identifiers with the node. . Note that the packet flow can be defined by, for example, any one of a final destination address, a final destination port, a source address, and a source port of the network layer, or any combination thereof.

The operation length of the Flow identifier is preferably fixed in the entire network. Further, a preferable value as the Flow identifier length is the MTU size when transmitting a normal datagram, and the value in the present embodiment.
MT for datagram transmission using MAC frame
In order to make the U size the same for any encapsulation, the encapsulation length is 2 octets, which is the encapsulation length used in MAC frames in Ethernet DIX format.

Assuming that the flow identifier length is 2 octets, the number of available cut-through identifiers is 65536 per MAC address (cut-through node identifier) newly introduced in the present embodiment. In other words, only by storing (learning) one MAC address for cut-through with an Ethernet switch or a bridge device with a learning function, 65536
Data link layer switching can be performed for each of the packet flows.

In consideration of the increase in the number of hosts and routers on the Internet, there is a possibility that 65536 cut-through identifiers may be insufficient. This can be dealt with by assigning a plurality of cut-through node identifiers to one device. Even in this case, the number of MAC addresses to be stored (learned) can be the number of assigned cut-through node identifiers.

FIG. 2 shows an example of the configuration of a router for performing high-speed transfer using the MAC frame format in the present embodiment. This router device has one or more network connection interfaces 211 to 21N, and is connected to networks 221 to 22N, respectively. The networks 221 to 22N are networks standardized by the IEEE802 committee, and include Ethernet (802.3) and token ring (802.5).

The datagram processing unit 201 performs network layer processing on a packet assembled from frames received by each of the network interfaces 211 to 21N. That is, it has a mechanism for determining a router or host to which a packet is to be transmitted next and a network connection interface based on destination information of a datagram such as IP. It is to be noted that the router device can be configured by omitting the packet forwarding function of the datagram processing unit.

The cut-through processing unit 202 performs protocol processing necessary for performing cut-through (data link layer switching) using the MAC frame format in the present embodiment. That is, Fl
By exchanging messages including ow identifiers and network layer level packet flow identification information, etc., a common understanding is obtained as to which cut-through identifier is used for cut-through transfer for a particular packet flow, and Based on this, the next-stage information table 261 described later is set. Also, the MAC address for cut-through is stored if necessary. The protocol used here is the FANP (Flow Attribute Notification Proto
col) and TDP (Tag Distribution Protocol).

Each of the network interfaces 211 to 21N
The configuration will be described using the network interface 211 as an example. The network interface 211 is
MAC control unit 231, MAC judgment unit 241, flow processing unit 251,
And the next-stage information table 261.

The MAC frame control unit 231 performs a MAC frame transmission / reception process.
Receives MAC frames and broadcast MAC frames addressed to the address. However, in order to receive a MAC frame or the like addressed to the group address,
Depending on the settings, other MAC frames can be received.

The MAC determination unit 241 determines whether or not the frame is a MAC frame having the MAC address format newly introduced in the present embodiment. If so (if cut-through transfer is to be performed), the MAC frame is received. The MAC frame is sent to the flow processing unit 251, and if not (normal network layer transfer is performed), the received MAC frame is sent to the datagram processing unit 201.

In the flow processing unit 251, the cut-through
By searching the next-stage information table 261 based on the MAC address information (only the cut-through node identifier may be used) and the Flow identifier existing in the MAC frame, the network connection interface to which the next frame is to be transmitted and the next-stage Information such as a cut-through identifier is obtained, the header information of the MAC frame is rewritten if necessary, and the MAC frame is directly transferred to the obtained network connection interface (without passing through the datagram processing unit 201).

FIG. 3 shows a datagram processing procedure of the present router when a cut-through MAC address is written in the destination MAC address area. For example, network interface 211 to network interface
Explain the datagram transfer to 21N.

The network interface 211 receives a MAC frame from the network 221. At this time, the MAC control unit 231 executes MAC frame processing (destination MAC address check, CRC check, etc.) and receives
If the destination MAC address of the MAC frame is stored as to be received by the router device, the MAC judgment unit
The frame is sent to 241 (S301). In other words, the destination MAC address is a normal MAC address of the own device, or the MAC address assigned to the own device for cut-through (cut-through vendor identifier + cut-through node identifier assigned to the own device) If so, the frame is passed to the MAC determination unit 241. Otherwise, the received frame is discarded (S302).

Prior to this, the cut-through processing unit 202
MAC assigned to own device for cut-through
The address is stored in the MAC control unit 231. Also,
If an Ethernet switch exists between the router and the adjacent node, the cut-through processing unit 202 of the router device sends the cut-through MAC address to the source when the cut-through MAC address is assigned to the own device. The frame written as the MAC address is transmitted to the adjacent node. By receiving this frame, the Ethernet switch on the way, the MAC for cut-through
It stores to which output interface the frame whose address is the destination MAC address should be forwarded.

Each node can store the cut-through MAC address of the adjacent node by receiving the frame. Alternatively, a procedure for exchanging the MAC address for cut-through between adjacent nodes may be separately defined by an inquiry / response type procedure or a unidirectional notification procedure.

Upon receiving the frame, the MAC determining unit 241 checks the vendor identifier of the destination MAC address (S30).
3) If the identifier is a cut-through vendor identifier, a MAC frame is sent to the flow processing unit 251 to perform cut-through transfer (S305); otherwise, the datagram processing unit 201 performs normal network layer transfer. (S304).

In the flow processing unit 251, the next stage information table
Referring to 261, an output network interface to be sent next and a cut-through identifier to be added next are obtained from the Flow identifier written in the MAC frame (S30).
Five). Here, the next stage information table 261 is searched by the Flow identifier, but may be searched by the cut-through identifier. In particular, when assigning a plurality of cut-through node identifiers to one node, it is preferable to search using the cut-through identifier.

If the corresponding entry does not exist when the flow processing unit 251 searches the next-stage information table 261, the MAC frame is discarded (S 306). Therefore, the source node is notified by the adjacent node that the adjacent node has set the next-stage information table based on the message exchange including the flow identifier and the packet flow identification information at the network layer level. Therefore, packets belonging to the packet flow are transmitted in the form of a MAC frame (131) including a cut-through identifier.
Until the notification, the packet belonging to the packet flow is transmitted in the form of a normal MAC frame (121 or 122) so that the adjacent node can perform transfer by normal datagram processing.

Although the message exchange for the cut-through process has been completed between the preceding node and the present router, the same message exchange has not been completed between the present router and the next node (or MA to be forwarded to the next node if no messages are exchanged)
In order to prevent the C frame from being discarded in S306, for example, there are the following three methods.

First, when message exchange with the preceding node is completed, an entry is created in the next-stage information table using the Flow identifier (or cut-through identifier) as a key, and the output interface column is blank. (Or write information indicating the datagram processing unit), and write only the upper layer protocol identification information such as the LLC header in the next-stage cut-through identifier column.
If the entry is in this state when the next-stage information table is searched in step (1), the MAC frame is passed to the datagram processing unit 201 (at this time, the upper layer protocol identification information is used).

Second, when the message exchange with the preceding node is completed, an entry is created in the next-stage information table using the Flow identifier (or cut-through identifier) as a key, and the Flowgram is processed by datagram processing. From the final destination network layer address of the packet flow specified by the identifier, the output interface to be sent next and the destination MAC address of the next node are obtained, and the obtained information and the necessary LLC header, protocol identifier, etc. This is a method of writing in the column of the column cut-through identifier. In this way, the MAC frame in which the corresponding entry exists in the next-stage information table proceeds to the steps after S307, and even if the message exchange with the next-stage node is not completed, the data link in the present router is performed. Layer switching transfer will be possible.

The third method is to write all data into the next-stage information table after the message exchange has been completed in both the previous stage and the next stage, and to pass the frame to the datagram processing unit 201 even when there is no entry. is there. When this method is used, for example, the cut-through node identifier 113
An address format that uses, for example, two octets of it for upper layer protocol identification information is newly adopted. Then, when receiving the cut-through MAC frame and sending it to the datagram processing unit 201, the protocol identification information described at a predetermined position in the cut-through node identifier may be used. I just need.

Alternatively, in order to prevent a frame including a cut-through vendor identifier from being transmitted in a state where the setting for cut-through transfer has been completed only to a node in the middle, the source node of the frame must be transmitted. Only after the end-end is notified that the message exchange and the setting of the next-stage information table are completed, the packet belonging to the packet flow is converted into a cut-through MAC frame and transmitted. Is converted to a normal frame and transmitted).

After the next stage information is obtained, the flow processing unit
At 251, the destination MAC address of the received frame and the Flow
The identifier is replaced with the acquired next-stage cut-through identifier (S307) and transmitted to the acquired output network interface 21N (S308).

Here, the destination MAC address and the flow identifier are replaced by each interface independently.
When the destination MAC address (MAC address for cut-through) and the flow identifier are made unique in all the networks so that the MAC address and the flow identifier can be managed, they need not be replaced because they have the same value.

The output network interface 21N may transfer the transmitted MAC frame as it is.
According to the above processing, the datagram can be transferred by the data link layer switching without performing the processing in the datagram processing unit 201.

FIG. 5 shows an example of the configuration of the next-stage information table for acquiring an information element to the next-stage router or host from the cut-through identifier. The entry of the cut-through identifier 501 may not be necessary depending on the search method, but the cut-through identifier serves as a search key. As described above, a Flow identifier may be used as a search key instead of a cut-through identifier. Output I /
F502 indicates a network interface to be provided next. The next-stage cut-through identifier 503 is used when the received MAC frame including the MAC address and the flow identifier for the cut-through is transmitted to the next router device or the destination host, and when it is necessary to replace the MAC frame header information or the like. Required information element.

At this time, the MA newly introduced in this embodiment is
In the case of a table entry for the next router or the final destination host that does not support the C frame (MAC frame for cut-through), the next-stage cut-through identifier 50
In 3, it is preferable to set a MAC address and encapsulation information (such as an LLC header) used for normal datagram transfer. In other words, MAC for cut-through
For a node that does not support the frame, the preceding node sends out a normal MAC frame using the normal MAC address and LLC header set in the next-stage information table. As a result, even when the router or host at the next stage does not support the MAC frame format in the present embodiment, cut-through is possible in the present router device. In addition, if the method of transmitting this normal MAC frame to the node that needs to forward the received frame to the upper layer processing is adopted, the MAC frame for cut-through does not include the upper layer protocol identification information. However, the node can operate without any problem.

When employing an address format in which, for example, two octets in the cut-through node identifier 113 are used for the upper layer protocol identification information, the received frame must be sent to the upper layer processing (and the upper frame processing must be performed). Sending a MAC frame for cut-through to a node (which supports the address format) allows the node to operate without any problem.

FIG. 4 shows a datagram processing procedure of the present router when a cut-through MAC address is written in the source MAC address area. As an example, datagram transfer from the network interface 211 to the network interface 21N will be described.

The network interface 211 receives a MAC frame from the network 221. At this time, the MAC control unit 231 performs MAC frame processing (destination MAC address check, CRC check, etc.)
If the address is the MAC address of this router,
The frame is sent to the determination unit 241 (S401). That is, the normal
Both the MAC frame and the cut-through MAC frame are sent to the MAC determination unit 241.

When a bridge device with a learning function is present in the network, the bridge device is connected to a normal MAC device.
If no distinction is made between the frame and the MAC frame for cut-through, if the cut-through MAC address assigned to the own device and sent by each node is entered as the source MAC address, the MAC address and output interface Will be memorized (learned).

Here, the transmission source node stores, in the transmission source MAC address area, the cut-through
It is assumed that the frame is transmitted with the MAC address entered,
Each node is for cut-through of adjacent nodes as described above
Since the MAC address can be known, the cut-through address assigned to the next node in the source MAC address area
You can also enter the MAC address and send the frame.

Upon receiving the frame, the MAC determining unit 241 checks the vendor identifier of the transmission source MAC address (S
402), if the identifier is a cut-through vendor identifier, a MAC frame is sent to the flow processing unit 251 to perform cut-through transfer (S404); otherwise, the datagram processing unit 201 performs normal network layer transfer. (S403).

In the flow processing unit 251, the next stage information table
Referring to 261, an output network interface to be sent next, a cut-through identifier to be added next, and a destination MAC address are obtained from the cut-through identifier written in the MAC frame (S404). Destination required here
The MAC address is the normal MAC address of the next node. In the case of the method of transmitting the frame by writing the MAC address for cut-through assigned to the next node in the source MAC address area, not the cut-through identifier,
The next stage information table 261 can be searched using the Flow identifier.

If the corresponding entry does not exist when the flow processing unit 251 searches the next-stage information table 261, the MAC frame is discarded (S 405). The method of preventing undesired discarding is the same as the case of writing the cut-through MAC address in the destination MAC address area described above.

After the next stage information is obtained, the flow processing unit
At 251, the source MAC address of the received frame and Fl
The ow identifier is replaced with the acquired next-stage cut-through identifier, the destination MAC address is also rewritten (S406), and sent to the acquired output network interface 21N (S40).
7).

The reason why the source MAC address and the flow identifier are replaced here is that the source MAC address and the flow identifier can be managed independently by each interface. ) And the Flow identifier are unique in all the networks, they do not need to be replaced because they have the same value. However, even in that case, the destination MAC address (normal MAC address) is replaced.

The sending network interface 21N may transfer the transmitted MAC frame as it is.
According to the above processing, the datagram can be transferred by the data link layer switching without performing the processing in the datagram processing unit 201.

FIG. 6 shows an example of the configuration of a next-stage information table for acquiring an information element to the next-stage router or host from the cut-through identifier. The entry of the cut-through identifier 601 may not be necessary depending on the search method, but the cut-through identifier serves as a search key. As described above, a Flow identifier may be used as a search key instead of a cut-through identifier. Output I /
F 602 indicates a network interface to be provided next. The next-stage cut-through identifier 604 is used to transmit the received MAC frame including the MAC address for cut-through and the flow identifier to the next router device or the destination host when it is necessary to replace the MAC frame header information or the like. Required information element. Next MA
The C address 603 indicates the normal MAC address of the router or host to be sent next, and
This is used when the MAC address format newly introduced in the present embodiment is used for the address.

The method for dealing with the case where the next router or host does not support the MAC frame format in the present embodiment or the case where the frame received by the node needs to be processed by the upper layer is described above. destination
This is the same as writing the MAC address for cut-through in the MAC address area.

In the above example, each node determines whether the received MAC frame should be used for cut-through transfer or normal datagram transfer based on the cut-through vendor identifier. The determination may be made based on the U / L bit instead of the through vendor identifier. That is, when the U / L bit is 1 (local address), the data to be cut-through transferred is 0,
(Universal address), it is determined that a normal datagram should be transferred. in this case,
The combination of the vendor identifier and the node identifier can be used for identifying a node in each network.

In the above-described procedure, a case where all interfaces (networks) are such as Ethernet, which transmits and receives MAC frames, has been described. For example, in FIG.
The present invention can be applied to a virtual connection type network such as M or frame relay. For example, in the case of an ATM network, the cut-through identifier is written in the VPI / VCI field of the cell header.

For example, if a MAC frame is received from the input interface and the output interface is ATM, the value of VPI / VCI is entered in the next cut-through identifier column of FIGS. You. As a result, the router assigns a predetermined VPI / VCI value to the output ATM interface only by referring to the cut-through identifier of the received MAC frame header without referring to header information such as the destination address of the packet. Packets can be transferred into cells. Here, the frame with the MAC header indicating the next-stage router may be cellized,
The packet from which the header has been removed may be converted into a cell (by adding an encapsulation header if necessary). The present invention is not limited to the above-described embodiment, and can be implemented with various modifications within the technical scope.

[0084]

As described in detail above, according to the present invention,
A virtual identifier that can be used for switching at the data link layer frame level so that there is no need to readjust the packet length, change it, or store many data link layer addresses in nodes or bridges, Ethernet switches, etc. Can be granted.

[Brief description of the drawings]

FIG. 1 is a diagram showing a normal MAC frame format and a MAC frame format used in the present embodiment.

FIG. 2 is a diagram showing a configuration example of a router that processes a MAC frame used in the embodiment.

FIG. 3 is a flowchart illustrating an operation example of a router when the MAC address format of the present embodiment is used for a destination MAC address.

FIG. 4 shows a MAC address of the present embodiment as a source MAC address.
9 is a flowchart illustrating an operation example of a router when an address format is used.

FIG. 5 is a diagram showing an example of a table configuration when the MAC address format according to the embodiment is used for a destination MAC address.

FIG. 6 shows the MAC address of the present embodiment as the source MAC address.
The figure which shows the example of a table structure at the time of using an address format.

[Explanation of symbols]

 201 Datagram processing unit 202 Cut-through processing unit 211-21N Network interface 221-22N Network (LAN) 231 MAC processing unit 241 MAC determination unit 251 Flow processing unit 261 Next stage information table

Claims (15)

[Claims] 1. A frame to be transferred at a data link layer level, wherein a packet flow to which the frame belongs is identified at a data link layer by using an area defined as an area for writing upper layer protocol identification information. Receiving means for receiving a frame in which information for writing is written, an identifier capable of identifying a packet flow, and address information of a next-stage node to which a frame belonging to this packet flow should be transferred. Storage means, by referring to the correspondence stored in the storage means as an identifier at least the information written in the area of the frame received from the preceding node by the reception means, without performing network layer processing At the data link layer level, the obtained address information of the next node is added to the frame and transferred. Router device characterized in that comprising a feeding means. 2. The router according to claim 1, wherein said receiving means includes means for receiving another frame to be transferred at a network layer level from a node belonging to a certain logical network, and said router apparatus assembles said another frame into a packet. 2. The router device according to claim 1, further comprising a network layer level transfer means for performing network layer processing on the packet, forming a frame again, and transferring the frame to a node belonging to another logical network. 3. A method for detecting, from the frame, information indicating whether or not address information written in an address area of the frame received by the receiving means includes an address allocated exclusively for data link layer level transfer. 3. The router device according to claim 2, further comprising means for transferring the frame to the data link layer level transfer means or the network layer level transfer means based on the detection result. 4. The storage means stores, as the address information or in addition to the address information, a new identifier which enables the next-stage node to identify the packet flow in a data link layer. The transfer means writes information based on the new identifier in an area defined as an area for writing upper layer protocol identification information of the frame. 2. The router device according to 1. 5. The storage means stores, as a part of the new identifier, a node identifier assigned to the next-stage node exclusively for transfer at the data link layer level. 5. The router device according to claim 4, wherein the node identifier is written in a destination address area or a transmission source address area of the frame. 6. The storage means stores, as a part of the new identifier, a node identifier assigned to the router device exclusively for the data link layer level transfer. 5. The router device according to claim 4, wherein the node identifier is written in a source address area of the frame. 7. The storage means further includes means for storing, as the address information or in addition to the address information, a correspondence relationship with protocol identification information capable of identifying an upper layer protocol of the packet flow, 2. The router device according to claim 1, wherein the transfer unit further includes a unit for writing the protocol identification information in an area defined as an area for writing upper layer protocol identification information of the frame. 8. The storage device according to claim 1, wherein the storage unit stores, as the address information, an address fixedly assigned to the next-stage node for the transfer at the network layer level. The router device according to 4, 7, or 8. 9. A packet to be transferred at a data link layer level, wherein a packet flow to which the frame belongs is identified at a data link layer using an area defined as an area for writing upper layer protocol identification information. Receiving means for receiving a frame in which information for writing is written, an identifier capable of identifying a packet flow, and address information of a next-stage node to which a frame belonging to this packet flow should be transferred. Storage means, by referring to the correspondence stored in the storage means as at least information written in the area of the frame received from the preceding node by the receiving means as an identifier, without performing the network layer processing The address information of the next node obtained at least to the destination address area of the frame. Router device characterized by comprising a transfer means of the data link layer level to transfer the frame Te. 10. A means for storing an address including a node identifier assigned to the router device exclusively for transfer at the data link layer level, and the receiving means writes the stored address in a destination address area. 10. The router device according to claim 9, further comprising: means for controlling to receive a frame. 11. A data link switch, which is present between the preceding node and performs frame routing using a destination address as a key, assigns a node identifier assigned to the router device exclusively for the data link layer level transfer. 11. The apparatus according to claim 10, further comprising: means for transmitting a frame in which the dedicated address is added to the transmission source address area to the preceding node so that the frame can be routed using the included address as a key. Router device. 12. A packet to be transferred at a data link layer level, wherein a packet flow to which the frame belongs is identified at a data link layer by using an area defined as an area for writing upper layer protocol identification information. Receiving means for receiving a frame in which information for writing is written, an identifier capable of identifying a packet flow, and address information of a next-stage node to which a frame belonging to this packet flow should be transferred. Storage means, and information written in a source address area of a frame received from a preceding node by the receiving means and an area defined as an area for writing the upper layer protocol identification information are used as identifiers in the storage means. The next stage obtained without performing the network layer processing by referring to the stored correspondence relationship The node address information, the router device characterized by by applying the destination address region and a transfer unit of the data link layer level transfer the frame of the frame. 13. The storage means stores, in addition to the address information, a new identifier that enables the next-stage node to identify the packet flow in a data link layer. 13. The apparatus according to claim 12, wherein the means further comprises means for writing the new identifier in an area defined as a source address area and an area for writing upper layer protocol identification information of the frame. Router device. 14. The storage means stores, as a part of the new identifier, a node identifier assigned to the next-stage node or the router device exclusively for the data link layer level transfer. 14. The router device according to claim 13, wherein: 15. An identifier for identifying a flow to which a packet received from a certain logical network belongs in a data link layer, and a next stage belonging to another logical network to which a data link layer frame of a packet belonging to the flow is to be transferred. If the information written in the address area of the received frame indicates that the frame should be transferred at the data link layer level, at least the The next node obtained without performing network layer processing by referring to the stored correspondence as an identifier, using information written in an area defined as an area for writing upper layer protocol identification information as an identifier. Wherein the address information is added to the frame and transferred.