JP2001086149A - Method for processing communication - Google Patents

Abstract

PROBLEM TO BE SOLVED: To speed up routing and packet transfer to easily extend a scale by routing-processing reception data from a communication port based on distributed relay route information in a routing accelerator and transferring it to another routing accelerator. SOLUTION: A router managing part 2 being a main processor provided with functions for managing a whole device, creating a routing table and executing distribution or the like is connected to a high speed router bus 1. The routing accelerators 3 having the function for high-speed routing are connected to the router bus 1 from one to eight modules. The router managing part 2 distributes the routing table to the routing accelerators 3, each routing accelerator 3 selects the route of reception packet data, that is, executes the routing processing and packet data is transferred to each routing accelerator 3 where a relay destination network is connected under it via the router bus 1.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a communication network system for connecting a plurality of networks, and more particularly to an internetwork apparatus called a router for connecting a plurality of networks at a network layer level and a plurality of networks at a data link layer level. The present invention relates to a communication processing method in an internetwork apparatus called a brouter, which has both a bridge function for connecting networks and a router function for connecting a plurality of networks at a network layer level.

[0002]

2. Description of the Related Art As a device for connecting a plurality of networks, a bridge for interconnecting in a data link layer (in particular, a media access sublayer) in a network system hierarchy, and an interconnecting in a network layer which is an upper layer thereof There are routers.

In a bridge, a MAC (Media Access Control) is used.
rol) manages the address and determines whether or not to relay a received frame (also referred to as received packet data) from one network to another network based on the contents of the destination MAC address and the relay control information in the received frame. Perform according to the filtering address table.

In a router, a predetermined route or an optimal route is selected according to an internetworking address in a received frame and an address resolution (route information) table in the router, and the received frame is relayed. .

[0005] There are several types of protocols used in the network layer.
(Internet Protocol) is well known. This IP protocol uses an IP address as an internetworking address. The address resolution table stores the MA of the adjacent route corresponding to the destination IP address.
The C address is described.

[0006] In recent years, devices having both a bridge function and a router function have appeared, and have been used for connecting heterogeneous networks. This device is called a brouter. The brouter routes frame data according to a protocol supported by the device, that is, a routable protocol among various protocols used in the network layer, and performs interconnection at the network layer. On the other hand,
For frame data that is not supported by the device, that is, according to a protocol that cannot be routed, relay processing of frame data in the media access sublayer, that is, bridge processing is performed.

As described above, a network device such as a bridge, a router device, and a brouter device has a configuration having at least two or more communication ports and a processor for performing the above-described relay processing called routing or bridging.

As a conventional example of such a configuration, a general operation of a router device will be described in more detail. By storing packet data received from a certain communication port in a buffer memory and performing routing processing on the packet data by a processor, the router device obtains a destination network. Then, by transmitting the packet data from another communication port corresponding to the destination network, the router device performs the routing of the packet data.

FIG. 2 is a diagram schematically showing a plurality of networks connected using a router. In this figure, 8
One network is connected by five routers. Looking at router A among these routers,
Networks A, B, C, and D are connected to ports a, b, c, and d of the router A, respectively. Now, suppose that data is to be sent from the terminal A1 in the network A to the terminal G2 in the network G connected to the router E.
The packet data addressed to the terminal G2 is transmitted with the physical destination address addressed to the port a of the router A. Router A
Receives this packet data, and transmits data to networks B, C,
D, ie, performs routing. In order to perform this routing, the router A performs the terminal G2
It is necessary to know in advance which of the networks B, C and D the packet data addressed to should be relayed. In this example, it is assumed that transmission is to be performed to the network B, that is, the router B, and the physical destination address of the packet data is transmitted to the port e of the router B. Similar routing processing is performed in the routers B and E, and the packet data finally reaches the terminal G2.

As a conventional technique for performing such a routing process at high speed, there is a technique described in Japanese Patent Application Laid-Open No. 62-181551. According to this, a buffer memory for storing packet data received from one port and a buffer memory for storing packet data received from the other port are provided separately and independently, and management of the buffer memory is performed by hardware. Is going.

As a technique for transferring packets between ports at high speed, Net One Systems Co., Ltd. 1990
Published on December 19, 1998, "TCP / IP and INTERNE
Technology described on page 17 of "T Overview". According to this, a high-speed bus is provided for high-speed port transfer.

[0012]

However, these conventional techniques have the following problems.

Although the router of the prior art can transfer packet data at high speed, the routing process is a bottleneck because there is only one routing means, and the number of ports that can be supported and communication traffic are limited.
Therefore, it is difficult to smoothly expand the configuration of the router port menu and the like from small to large, and to improve the performance according to the traffic and the number of ports.

Further, the network configuration and the like are recognized during the operation of the network, and relay information for routing processing (information for relaying that each router knows in advance in the above-described conventional example) is dynamically obtained. In recent years, the need for dynamic routing that generates, adds, changes, and deletes has been increasing. That is, a routing protocol (for example, RI in the TCP / IP protocol group) for exchanging information about the network between the routers.
P: Routing Information Protocol or OSPF: Open S
hortest Path First). Further, a network management protocol for communicating management information such as performance information of the router itself with a management master station on a network (for example, SNMP: Simple Network Management Protocol in the TCP / IP protocol group).
In the processing of ocol, etc., the means for performing routing in the conventional example must also be used, so that the original relay performance cannot be sufficiently exhibited.

Therefore, in the conventional router, a high-speed LAN of 100 Mbps (Mega bit per second) which has recently emerged.
(Local Area Network) FDDI (Fiber Distribut
edData Interface), broadband ISDN (hereinafter abbreviated as B-ISDN), which is expected to be widely used in the future, and ATM (Asyncronized Transfer Mode), which is one form thereof
It is difficult to support a high-speed line of 55 Mbps.

An object of the present invention is to provide an internetwork apparatus and a communication processing method which can perform routing and packet transfer at high speed, and can easily expand the scale and improve the performance.

A further object of the present invention is to provide an internetwork apparatus and a communication processing method that take into account data transfer during bridging processing and routing processing when the above-described function as a brouter is provided.

A further object of the present invention is to provide a router device and a communication processing method capable of connecting to an existing device in order to increase the types of protocols that can be processed.

[0019]

In order to achieve the above object, an internetwork apparatus according to the present invention has one or more routing accelerators for performing routing by assisting a main processor operating as a host management unit, A communication port through which each of the routing accelerators transmits / receives data to / from a network. The communication processing method according to the present invention is characterized in that the host management unit generates / changes relay route information for data relay. Distributing the relay route information to the plurality of routing accelerators, and in each of the routing accelerators, routes received data from the communication port based on the distributed relay route information,
It is characterized in that it is forwarded to another routing accelerator.

Further, in order to achieve the above object, the internetwork apparatus of the present invention has first connection means for connecting between routing accelerators, and the first connection means includes the entire internetwork apparatus. The main processor is connected as a managing unit for managing. Then, there is provided second coupling means for coupling the plurality of communication port units for transmitting and receiving data to and from the network and the routing accelerator independently of other routing accelerators.

Further, in order to achieve the above object, the internetwork apparatus of the present invention is provided with third coupling means for coupling between the main processor and the routing accelerator independently of the first coupling means.

Further, in order to achieve the above object, the internetwork apparatus of the present invention has an auxiliary processor in addition to the main processor and the routing accelerator.
Then, the auxiliary processor and the routing accelerator are connected by the first connecting means or the third connecting means. As functions of the auxiliary processor, a broadcast transfer function for bridge relay, an interface function with other devices, or a processing function of a routing protocol that the main processor cannot process is provided for the function expansion.

Further, in order to achieve the above object, an internetwork apparatus of the present invention provides a routing accelerator with a buffer for storing data as a router circuit, and a filtering under predetermined conditions in parallel with the storing operation. A filtering assist unit for extracting necessary information of the received packet data and searching the relay route selection information table.

[0024]

The routing accelerator performs a routing process on the received packet data and transfers the data to another routing accelerator as needed. That is, the type of the data frame is determined, and if the data frame is recognized as a data frame according to a protocol that cannot perform routing relay processing, the data is transferred to a plurality of other routing accelerators for bridge relay.

After generating / changing a routing table for router relay, the management unit distributes this routing table to all the routing accelerators. Each routing accelerator performs a routing process on the received packet data from the communication port unit connected by the second coupling unit with reference to the distributed routing table, and uses the first coupling unit as necessary. To transfer data to another routing accelerator. That is, the type of the data frame is determined,
When it is recognized that the data frame conforms to the protocol that cannot perform the routing relay processing, the data is transferred to a plurality of other routing accelerators in order to perform the bridge relay. The routing accelerator to which the data has been transferred further transmits the packet data to the network using the communication port unit connected by the second coupling unit. Therefore, the internetwork apparatus of the present invention can perform distributed routing / packet relaying by a plurality of sets of the routing accelerator and the communication port unit.

A management frame addressed to the main processor for the main processor to generate / change the routing table and to communicate with the network management master is transferred from the routing accelerator to the main processor using the first coupling means. Is done. In addition, the routing accelerator receives data received from the communication port unit connected by the second coupling means, determines the type of the data frame, and is a data frame according to a protocol that cannot perform routing relay processing. When it is recognized, it can be transferred to the main processor. If the main processor can process the protocol, the main processor performs a routing process, retransmits the data frame to an appropriate destination routing accelerator, and performs a routing relay of the packet data according to the special protocol.

The third coupling means performs an operation of distributing the routing table to all the routing accelerators by the main processor and an operation of transferring a management frame addressed to the main processor from the routing accelerator to the main processor.

The auxiliary processor has a broadcast transfer function for bridge relay according to the function implemented in the auxiliary processor,
In correspondence with an interface function with another device, a processing function of a routing protocol that cannot be processed by the main processor, and the like, the processing of the main processor or a bridge relay function of the destination routing accelerator is assisted. For example, when the auxiliary processor assists the broadcast transfer function for bridge relay, when the routing accelerator receives a data frame according to a protocol that cannot perform routing relay processing, the frame data to be bridge-relayed to this auxiliary processor. To transfer. After temporarily storing the transferred frame data in the buffer memory, the auxiliary processor transfers the data to all the other routing accelerators using the first coupling unit.

In the routing accelerator, in parallel with storing the received packet data in the buffer, the filtering assist means performs filtering under predetermined conditions, and discards the packet data that does not need to be relayed. For received packet data that requires relaying,
The processor and the routing assist unit determine the destination, and transfer the packet to the routing accelerator serving as the destination.

[0030]

An embodiment of the present invention will be described below with reference to the drawings. Although the internetwork apparatus of the present invention has the function of the above-mentioned brouter apparatus, in the following description, the main purpose of the routing processing will be mainly described, and hence the apparatus will also be referred to as a router.

1. First, the overall configuration of a router according to an embodiment of the present invention will be described with reference to FIG. The router of this embodiment has a plurality of modules for performing routing, and the performance can be easily improved by adding modules.
Each module has one routing accelerator, and each routing accelerator further has one or more communication ports. FIG. 1 is an overall block diagram of a router. In FIG. 1, 1 is 200 MBytes, which is the first combining means.
It is a high-speed router bus with a throughput of 1 / sec. The router bus 1 is connected to a router management unit 2 which is a main processor having a function of managing the entire apparatus and a function of generating and distributing a routing table. Further, a routing accelerator 3 having a function of performing high-speed routing can be connected to the router bus 1 from 1 to 8 modules. The router management unit 2 distributes a routing table to the routing accelerators 3, and each of the routing accelerators 3 selects a route of the received packet data, that is, performs a routing process based on the routing table, and connects the relay destination network to the subordinate. The packet data is transferred to the routing accelerator 3 via the router bus 1.

Further, each routing accelerator 3
Below, as shown in FIG. 4, as a bus which is the second coupling means, for example, a generally known 33 MBytes /
EISA bus (Extended Industry Standard Archit)
ecture Bus) 4. Various communication control units 51 to 56 (hereinafter also referred to as ports) are connected to this bus.
Here, only one high-speed communication port 51 of the FDDI of 100 Mbps class, which is a communication port corresponding to the throughput of the EISA bus, is connected to the routing accelerator 3 on a one-to-one basis. Similarly, only one communication port 52 of 155 Mbps ATM (B-ISDN), which is a high-speed line, is connected to the routing accelerator 3. Also, for example, a communication port 53, 4M of Ethernet (registered trademark) which is a LAN of 10 Mbps.
Token which is LAN of bps to 16Mbps
Ring LAN communication port 54 or 1.5 Mb
A plurality of communication ports for medium / low speed communication such as the N-ISDN communication port 55, which is a line of ps, can be handled by the capacity of one routing accelerator 3 and the throughput of one EISA bus. are doing. Note that the set of the routing accelerator 3 and the communication ports 51 to 56 may be mounted on a separate substrate or may be mounted on the same substrate.

2. Routing Routing in a router having the above-described basic configuration will be described below.

First, it is assumed that the management unit 2 distributes a routing table to all the routing accelerators 3 and that each of the routing accelerators 3 has routing information. For example, the management unit 2 collects routing information by exchanging with another router using the above-described routing protocol such as RIP or OSPF, or by a user statically setting in advance.

This routing table is a set of routing information indicating one piece of routing information for each network address. Further, for a so-called sub-network that divides a network address and operates as a plurality of small-scale networks, routing information is defined for each sub-network.

FIG. 14 shows an example of one piece of routing information. In this example, only the IP protocol is targeted, and additional information such as the type of network service is omitted. In FIG. 14, the routing table 400
Is a field 401 of an IP address indicating the destination network, subnet mask data 402 indicating subnet information of the destination network, and an IP address 4 of the next router.
03, sending interface 4 for relaying to the next router
04 and a pointer 405 to the next entry.

In this embodiment, the transmission interface 404 has an identification number (RA number) of a transfer destination RA (routing accelerator) and an identification number (LC number) of a port in the transfer destination RA to be transmitted. The pointer 405 to the next entry is a field necessary for structuring the routing information inside the RA, but has no meaning when distributed from the above-mentioned RM (router management unit).

The routing table, which is an aggregate of the routing information 400 in the RA, has a structure shown in FIG. 15 for high-speed retrieval. Hash function 410
Projects certain data onto a smaller amount of data. In this embodiment, for example, 24 bits of the network address portion of the IP address are projected to 8 bits. Since the output of the hash function is 8 bits, the number of entries in the hash entry table 420 is 256. Each entry in the hash entry table 420 stores a pointer to the routing information 400 corresponding to the result of hashing the network address. Since the hash compresses the amount of data, different network address values may have the same 8-bit address value as a result of hashing. Therefore,
As shown in FIG. 15, the routing information 400 having a network address that has the same result by hashing
Are connected by a pointer 405.

The structure for searching the routing table corresponding to the IP address and the structure of the routing table described above are described in Douglas E. Come.
r “Internetworking with TCP / IP VOLUME II” (81-
(P. 84), the present embodiment also employs a routing table and its data structure in the RA according to this method.

FIG. 3 is a diagram showing various routing cases in the router of the present invention having a plurality of modules / ports. In FIG. 3, the management unit 2
Is abbreviated as RM, the routing accelerator 3 is abbreviated as RA, and the communication ports 51 to 56 are abbreviated as LC.

Case 1 shows a case of disposal. The routing accelerator RA (A) temporarily stores a packet received by the communication port LC (a) under its control in a packet buffer, and performs filtering and routing. Here, as a result of the filtering / routing, the destination of the packet is the communication port LC just received.
If the packet is in the direction of (a), the packet is discarded.

Case 2 shows a case in which routing is performed between ports in the routing accelerator RA (A). The routing accelerator RA (A)
The packet received by the communication port LC (a) under its own control is temporarily stored in a packet buffer, and filtering and routing are performed. Here, as a result of performing the filtering / routing, the destination of the packet received by the communication port LC (a) is set to the own routing accelerator RA.
If the received packet is at the communication port LC (b) under the control of (A), the received packet is returned, transferred to the communication port LC (b), and relayed. In this case, only one routing accelerator RA (A) functions as a router.

Case 3 shows a case where routing is performed between two routing accelerators RA (A) and (B) via a router bus. The routing accelerator RA (A) has a communication port LC under its control.
(A) temporarily stores the received packet in a packet buffer, and performs filtering and routing. Here, as a result of the routing accelerator RA (A) checking the destination of the received packet, if it is the communication port LC (d) under the control of the other routing accelerator RA (B), the routing accelerator RA (A) and the routing accelerator RA (A) Information for packet transfer is exchanged between accelerators RA (B), and packets are transferred from routing accelerator RA (A) to routing accelerator RA (A).
Transferred to (B).

Case 4 shows a case where the received packet is a packet addressed to the own router.

The packet addressed to the own router is the aforementioned routing protocol frame (RIP or OSPF frame) or frame data of the network management protocol SNMP. The routing accelerator RA (A) temporarily stores a packet received by the communication port LC (a) under its control in a packet buffer, and performs filtering and routing. Here, for example, if the destination IP address of the IP packet is its own IP address, the received packet is transferred to the management unit 2.

3. Configuration of Routing Accelerator Next, the configuration of the routing accelerator 3 will be described.

FIG. 4 is a block diagram of the routing accelerator 3. The routing accelerator 3 includes a filtering assist unit 302 for filtering received packets by hardware, a lower processor 306 mainly responsible for filtering, and a lower processor 30.
6, local memory 305, routing assist unit 3 for routing received packets by hardware
08, upper processor 31 mainly responsible for routing
0, local memory 309 for upper processor 310, filtering assist unit 302, and lower processor 306
And a command descriptor buffer 303 for exchanging information with the lower processor 306 and the upper processor 310, and converting a network address (eg, an IP address) into a physical address (eg, a MAC address). Address conversion assist unit 304, interface unit 301 with EISA bus 4,
Packet buffer 307 and router bus transfer control unit 3
11.

First, the routing process will be described with reference to FIG. 16 along an upward data path for transferring data from the communication port 5 to the router bus 1 via the routing accelerator 3.

All the packets received by the communication port 5 are EI by the DMA transfer performed by each communication port 5.
The data is temporarily stored in the packet buffer 307 via the SA bus interface unit 301 (1601). In parallel with the operation of transferring the packet to the packet buffer 307, the filtering assist unit 302 extracts only the leading part of the packet necessary for filtering, performs comparison and determination under predetermined conditions, and performs filtering. In the present embodiment, the type of protocol to be handled, such as TCP / IP,
Alternatively, it recognizes one of the OSIs, and in the case of other protocol types, performs only filtering as a bridge without performing subsequent filtering / routing (1602). In addition, filtering can be performed by comparing with a template which is an address data pattern corresponding to an application set by the user.

The filtering assist unit 302 stores the result in the command descriptor buffer 303 in association with the pointer to the packet buffer 307 of the packet. The lower processor 306 takes over the result, continues the further filtering by software, and discards the packet if it determines that the packet is unnecessary (1603). The lower processor 306 also performs buffer management such as monitoring the free buffer status in the upstream direction. If the packet is not discarded, the routing instruction is sent to the command descriptor buffer 303.
Is stored in association with the pointer to the packet buffer 307. This completes the filtering process, and thereafter, the routing process is performed as follows (1604), and the packet data is transmitted to the router bus (1605).

The upstream IP routing process will be described with reference to FIGS. FIG. 17 is a flowchart showing the upstream IP routing process of the upper processor 310. First, the upper processor 310
Receives the routing instruction from the lower processor 306, reads the received packet data with reference to the pointer to the packet buffer stored in the command descriptor buffer 303. Here, it is assumed that the received packet data is an IP packet.

In FIG. 17, first, the destination IP address of the received packet is read out, the network portion of the IP address is subjected to the hash function processing shown in FIG. 15, and the entry pointer of the routing table is obtained (176).
0). Next, the destination IP address (tentatively, a) in the routing information indicated by the entry pointer is obtained (1).
761). Since the destination IP address (32 bits) of the received packet may use the subnet, the network address value (probably b and b) is obtained by deleting the host address part (lower n bits) using the subnet mask in the routing information. Is obtained (1762). And
a and b are compared (1763). If they match, the routing information is determined to be information for managing the destination of the received packet, and the IP address 403 of the next router in the routing information and the transmission interface 404 are determined. (1
764). If a and b do not match, the routing table is management information of another network address having the same value in the hash processing, so that the pointer 405 to the next entry in the routing information is read (1765). The same processing is repeated for the routing information indicated by the pointer.

Incidentally, depending on the received IP packet, further optional processing may be required.

When the IP processing is completed, the upper processor 310 activates the router bus transfer control unit 311 to transfer the packet to the destination module, specifically, the routing assist unit 308 of the destination. As a transfer operation, the transfer source router bus transfer control unit 311 notifies the transfer destination router bus transfer control unit 311 of command data such as a transfer request, a next hop IP address, a destination port number, and the like in advance, and transmits the transfer destination host processor. 310 receives this. When the upper processor 310 of the transfer destination is ready to accept the transfer, by returning a reply to the router bus transfer controller 311 of the transfer source,
Start packet transfer.

The above is the processing in the upstream direction by the routing accelerator 3 of the transfer source.

Next, processing performed by the routing accelerator 3 at the transfer destination along the flow of packets in the downward direction (from the router bus to the communication port) will be described with reference to FIG.

When the router bus transfer control unit 311 receives a transfer request from the transfer source router bus transfer control unit 311,
The upper processor 310 is notified of this fact. The host processor 310 performs downstream buffer management. The host processor 310 gives a storage destination start pointer to the packet buffer 307 to the router bus transfer control unit 311. The router bus transfer control unit 311 stores the packet received from the router bus 1 in the packet buffer 307 while updating the buffer address while performing handshaking with the transfer source router bus transfer control unit 311. After storing the packet, the upper processor 310 stores the storage end notification, the destination port number, and the pointer to the buffer in the command descriptor buffer 303 (1801).

The lower processor 306 receives this and performs the following processing. The lower processor 306 sends the next hop I
The P address is input to the address conversion assist unit 304.
On the other hand, the address conversion assisting unit 304 converts the network address (IP address) into a physical address (MAC address) and informs the lower processor 306 (18).
02). Note that this IP address to MAC address conversion method is described in ARP protocol (Address Res
It is generally known that this is possible with the solution protocol. Another method is to use IP address and MA.
A conversion table with the C address may be provided in advance, and the conversion process may be performed using the conversion table. Alternatively, the conversion process may be performed by constructing a learning type conversion table for registering the result of the ARP protocol.

Further, the lower processor 306 divides the packet in the packet buffer 307 if it is necessary to divide (segment) the packet data for adjusting the packet length corresponding to the destination communication port (180).
3).

Thereafter, the lower processor 306 adds a destination physical address to the packet data (or a plurality of segmented packet data) to form a packet format of a corresponding port. In this way, when the lower-level processor 306 is ready for the transmission packet, it activates the subordinate communication port 5 to perform transmission to the network (1804).

The router according to the present embodiment can support multiple protocols.

That is, in the above-described filtering processing, the TCP / IP protocol and the OSI protocol are determined from the received packet header. Based on the result, IP routing or OSI routing is continued. If the protocol cannot be determined, the packet is routed as a bridge. The bridge relay function will be described later.

In this embodiment, filtering is performed by the filtering assist unit 302 and the lower processor 306.
In this configuration, routing is performed by the upper processor 310 and the routing assist unit 308, and address conversion is performed by the lower processor 306 and the address conversion assist unit 304. The processor may perform the processing by software.

4. Extensibility Next, scalability, which is a feature of the present invention.
Will be described. As described above, in the configuration of the router according to the present invention, a small-scale router can be configured by a module including a low-speed line having eight or less communication ports, one routing accelerator, and a module of the management unit. If a communication port and a routing accelerator are paired to add a module, a middle- to large-scale router can be realized with the same architecture as a small-scale router.

5. Separation of Packet Data Transfer Bus and Control Data Transfer Bus Next, another embodiment for further increasing the speed of the above-described embodiment will be described with reference to FIG. In the present embodiment, the router bus 1 is dedicated to the transfer of packet data between the routing accelerators 8, and the transfer of control data such as management frames and relay information between the router management unit 2 and the routing accelerator 8 is performed by the third method. Another bus as a coupling means, that is, a control system bus 5 was provided. Accordingly, a pair of routing accelerators having packet data to be relayed by a router can perform packet data transfer using the router bus 1 even if another routing accelerator is transferring data with the router management unit 2.
Routers are faster as devices. In order to connect the control bus 5 to the routing accelerator 8, the bus interface of the router bus transfer control 311 in FIG. 4 is replaced by a two-channel DMA control circuit corresponding to the router bus 1 and the control bus 5. realizable.

Further, the scalability of the router system when the third coupling means is used will be described with reference to FIGS. According to this embodiment, the performance of the router can be improved and a large-scale network can be accommodated by connecting the routing accelerator of the present invention to the conventional router via a bus.

FIG. 6 shows an embodiment of a router in which the conventional device is extended. The portion surrounded by the dotted line in FIG. 6 is a component of the conventional router.

FIG. 7 is a block diagram showing only the components of the conventional router. In FIG. 7, reference numeral 5 denotes a bus that is used as a standard in routers and small information processing devices, and is a VME bus here. For the VME bus 5, a main processor unit 6 including a main memory and a microprocessor, etc., a line control unit 71 as a communication port unit,
LAN controllers 72 and 73 are connected. Normally, this conventional router performs routing as follows.

For example, the main processor unit 6 receives all packets received by the LAN control unit 72 and selects a destination port. Then, for example, if the destination port is the LAN control unit 73, routing relay is performed by transferring the packet data to the LAN control unit 73 via the VME bus 5. Therefore, the VME bus 5 is used for exchanging both packet data and control information between the main processor unit 6 and each of the communication control units 71 to 73.

Returning to FIG. 6, description of the embodiment of the present invention will be continued. Here, the routing accelerator 8 can make a third bus connection in addition to the first and second buses described above.
The third bus is connected to the main processor unit 6 by using a VME bus used in a conventional router. On the other hand, the high-speed router bus 1 transfers only packet data between the routing accelerators 8.
That is, in this embodiment, the main processor unit 6 manages the entire router, and the router management unit 2 shown in FIG. 5 is substituted for the function such as distributing the relay information table to each of the routing accelerators 8. The conventional router can be extended to the router of the present invention having both high speed and expandability.

In this embodiment, a router is used as a conventional device. However, a small information processing device such as a workstation and a routing accelerator 8 are used instead of a dedicated router.
The connection also provides workstation-specific scalability, such as early response to new protocols.

6. Function Expansion by Adding Auxiliary Processor Next, another embodiment for further improving the function expandability of the router of the present invention will be described with reference to FIG. 8 has an auxiliary processor 7 in addition to the router shown in the block diagram of FIG. This auxiliary processor 7
May share a part of the processing of the router management unit 2, that is, the function of managing the entire apparatus, or the processing of the routing protocol and the function of creating the relay route selection information table. However, here, in order to improve the function expandability of the router, an example of realizing an interface function for connecting to another information processing device, that is, a proxy function using the third bus described above, is described. explain.

In FIG. 9, the router 100 is equipped with an external interface board 101 for connecting an external connection bus (eg, SCSI) 200 and the router bus 1 as an auxiliary processor. The information processing device 200 such as a workstation is provided with a SCSI bus 20 as an external connection bus.
One. Here, it is assumed that the information processing device 200 can process a protocol that cannot be processed by the router 100, for example, a connection-oriented conventional host-terminal communication protocol or the like. When receiving the packet data according to the conventional protocol from the line control port 56, the router 100 transfers the received packet data from the routing accelerator 3 connected to the line control port 56 to the external interface board 101 via the router bus 1. . Further, the external interface board 101 performs the relay processing of the received packet data by transferring the received packet data to the information processing device 200 via the SCSI bus 201.

FIG. 10 shows the external interface board 10.
1 is a block diagram showing a configuration example of FIG. The external interface board 101 also enables a logical interface between the routing accelerator 3 and the external information processing device 200. Therefore, an interface processor 102 is provided, and a local memory 103 for operating the processor 102 is provided. Further, the external interface board 101 includes a router bus transfer control unit 104, S
A CSI bus transfer control unit 105 includes a packet buffer 107 for storing packet data and a control unit 106 for the packet buffer 107. Then, the processor 102 and the packet buffer 10
The bus switch 108 is provided so that external access to the PC 7 can be performed simultaneously. Therefore, the operation of the external interface board 101 is as follows. The router bus transfer control unit 104 stores the packet data from the packet bus 1 in the packet buffer 107. The processor 102 adds a descriptor for accessing the packet data by the external information processing device 200 as necessary, and notifies the external information processing device 200 of the packet data. The external information processing device 200 reads the packet data from the packet buffer 107 via the SCSI bus transfer control unit 105.

As described above, since the router has the auxiliary processor having an external interface function, the function can be expanded.

7. Next, a description will be given of an embodiment of a bridge relay process when a routing process is impossible in the Internet device according to the present invention.

If the received packet data conforms to a protocol that cannot be processed by the routing accelerator, as described above, the routing accelerator is connected to another router or information processing device, and the processing by the other device is performed. If expected, the routing accelerator can process the received packet data by transferring it to another router or information processing device. The same applies to the case where processing by the above-described auxiliary processor and main processor can be expected. However, since there are various communication protocols, a case where packet data which cannot be routed even by the entire internetwork apparatus of the present invention may be received naturally occurs. In this case, bridge relay processing is performed for all possible networks by performing only conversion at the data link level and relaying. That is, the routing accelerator performs bridge relay by broadcasting to a plurality of other routing accelerators.
However, in the present internetwork apparatus, a case where router relay and bridge relay are mixed may naturally occur. In this case, the broadcast transfer to the router bus takes time, and the relay performance as a device may be reduced. This will be described with reference to FIG.

In the internetwork apparatus of FIG. 11, the routing accelerator RA (2) performs a routing process on the received packet of the communication port LC (2) under the control thereof, and as a result, the other routing accelerator RA (3)
A routing accelerator RA (3) via a router bus for relaying to a communication port LC (3) of the router
It is assumed that transfer of packet data has started (t in FIG. 11).
1).

On the other hand, the routing accelerator RA
As for (1), as a result of trying to route the received packet of the communication port LC (1) under the control, the routing accelerator RA (1) judges that the packet data cannot be routed, and bridges all other routing accelerators. Attempts to transfer packet data by broadcast to relay. However, at this point, the routing accelerator RA (3) is not transferring data from the routing accelerator RA (2), so that the routing accelerator RA (1) cannot transfer data by broadcasting. Therefore, the routing accelerator RA (1)
It waits until A (3) is ready to receive data from the router bus (incomplete data transfer indicated by t2 in FIG. 11).

Further, the routing accelerator RA
It is assumed that before the data transfer from (2) to routing accelerator RA (3) is completed, data transfer for router relay from another routing accelerator RA (4) to routing accelerator RA (5) occurs. (The data transfer indicated by t3 in FIG. 11), the routing accelerator RA (1) waits until the routing accelerator RA (5) is ready to receive data from the router bus.

In other words, a routing accelerator that intends to perform a broadcast transfer to all the other routing accelerators for bridge relaying waits for the completion of data transfer between the other routing accelerators, so that routing from a subordinate communication port is possible. It is necessary to keep storing the received data. As a result, the routing processing performance of an internetwork device connected to a network that causes a relatively large number of bridge relays is reduced.

In order to solve such a problem, in the present invention, the above-mentioned auxiliary processor is used as an auxiliary function of bridge relay. This embodiment will be described with reference to FIGS.

In the internetwork apparatus shown in FIG. 12, it is assumed that data transfer occurs from routing accelerator RA (2) to routing accelerator RA (3) (data transfer indicated by t1 in FIG. 12). Here, the routing accelerator RA (1) corresponds to FIG.
Similarly to the above, it is determined that the packet relay from the subordinate communication port should be bridge-relayed. At this time, the routing accelerator RA (1) performs packet data transfer to the bridge relay assist unit 700 instead of performing broadcast transfer to all other routing accelerators (indicated by t2 'in FIG. 12). Data transfer). After that, the bridge relay assist unit 700
Attempts to broadcast to all routing accelerators. Here, similarly, from the routing accelerator RA (4) to the routing accelerator RA (4).
Even if data transfer to (5) occurs (data transfer indicated by t3 in FIG. 12), the bridge relay assist unit 70
0 only causes a wait for the broadcast transfer,
RA (1) does not need to keep accumulating subsequent routable received data. Therefore, it is possible to prevent a decrease in routing processing performance.

Next, the configuration of the bridge relay assist unit 700 will be described with reference to FIG. The bridge relay assist unit 700 mainly includes a buffer management processor 701 for performing buffer management, a local memory 702 for storing programs and variables of the processor 701, and a packet buffer 703 for temporarily storing data to be bridge-relayed. And a transfer control circuit 704 for connection with the router bus, and a bus switch circuit 705 for controlling disconnection of the bus for simultaneously transferring data from the router bus and accessing the local memory 702 of the processor 701. As described above, the packet data to be bridge-relayed from the routing accelerator is stored in the packet buffer 703 via the router bus transfer control circuit 704. At this time, the queue of the packet buffer 703 is managed by the buffer management processor 701, and the buffer management processor 701 performs the broadcast transfer to all the routing accelerators in the order stored in the packet buffer 703.

As described above, according to the present embodiment, the bridge processing is assisted by using the auxiliary processor, so that the router processing performance in the brouter can be improved.

[0086]

As described above, the present invention has the following effects.

Since the management unit for performing non-routing processing such as a network management function and the routing accelerator for performing routing processing are made independent, the speed can be increased because the routing accelerator is dedicated to the routing processing.

Since a plurality of packet data can be simultaneously routed by a plurality of routing accelerators, the routing performance of the entire apparatus becomes faster.
Also, you can add more routing accelerators,
The configuration of the router can be easily expanded according to the required number of networks and communication traffic.

Further, by using the auxiliary processor of the present invention, it is possible to perform bridge relay processing at high speed, and to extend the function by connecting to an external device.

Therefore, according to the present invention, an internetwork apparatus having expandability and performing high-speed router processing and bridge processing can be realized.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an internetwork device according to an embodiment of the present invention.

FIG. 2 is a diagram schematically illustrating a network system using a router.

FIG. 3 is a diagram showing a case of routing relay when one embodiment of the present invention is used.

FIG. 4 is a block diagram illustrating a routing accelerator according to an embodiment of the present invention.

FIG. 5 is a block diagram showing a configuration of an internetwork apparatus according to an embodiment of the present invention in which a packet dedicated bus and a control data bus are separated.

FIG. 6 is a block diagram showing a configuration of an internetwork device in which a routing accelerator according to one embodiment of the present invention and a conventional router device are combined.

FIG. 7 is a block diagram illustrating a configuration of a conventional router device.

FIG. 8 is a block diagram showing a configuration of an embodiment of the present invention provided with an auxiliary processor.

FIG. 9 is a block diagram showing a configuration of an internetwork apparatus in which an embodiment of the present invention and an information processing apparatus are connected by an external bus.

FIG. 10 is a block diagram showing a configuration of an external interface board according to one embodiment of the present invention.

FIG. 11 is a diagram showing a case of bridge relay without a bridge relay assist unit in one embodiment of the present invention.

FIG. 12 is a diagram illustrating a case of bridge relay with a bridge relay assist unit according to an embodiment of the present invention.

FIG. 13 is a block diagram illustrating a configuration of a bridge relay assist unit according to an embodiment of the present invention.

FIG. 14 is a diagram showing a configuration of routing information according to one embodiment of the present invention.

FIG. 15 is a diagram showing a configuration of a routing table according to an embodiment of the present invention.

FIG. 16 is a flowchart illustrating an uplink routing process according to an embodiment of the present invention.

FIG. 17 is a flowchart showing an upstream IP routing process according to an embodiment of the present invention.

FIG. 18 is a flowchart illustrating a downlink routing process according to an embodiment of the present invention.

[Explanation of symbols]

REFERENCE SIGNS LIST 1 router bus 2 router management unit 3 routing accelerator 4 lower bus 5 control bus
7 ... Auxiliary processor, 51-56 ... Communication control unit, 301
... EISA bus interface unit, 302 ... Filtering assist unit, 303 ... Command descriptor buffer, 304 ... Address conversion assist unit, 305 ... Local memory, 306 ... Lower processor, 307 ... Packet buffer, 308 ... Routing assist unit, 3
09: local memory, 310: upper processor, 31
1. Router bus control unit.

 ──────────────────────────────────────────────────続 き Continuing from the front page (72) Inventor Hiromichi Enomoto 1 Horiyamashita, Hadano-shi, Kanagawa Pref. Inside the Kanagawa Plant of Hitachi Ltd. Inside the Kanagawa Plant (72) Inventor Osamu Takada 1099 Ozenji Temple, Aso-ku, Kawasaki-shi, Kanagawa Prefecture, Ltd.System Development Laboratory, Hitachi, Ltd.

Claims (7)

[Claims] 1. A host management unit, a plurality of routing accelerators each transferring data between the host management unit and another routing accelerator, and each of the routing accelerators transmits data to and from a network. A communication processing method in an apparatus having a communication port for transmitting and receiving data, wherein the host management unit generates / changes relay route information for data relay, and transmits the relay route information to the plurality of routing accelerators. A communication processing method, comprising: distributing, in each of the routing accelerators, routing processing of data received from the communication port based on the distributed relay path information, and transferring the received data to another routing accelerator. 2. A host management unit, a plurality of routing accelerators each of which mutually transfers data between the host management unit and another routing accelerator, and each of the routing accelerators transmits data between the network and a network. A communication processing method in a device including a communication port for transmitting and receiving the data, wherein the host management unit creates a routing table for routing the data, and the routing table is transmitted to the plurality of routing accelerators. Distributing, and in each of the routing accelerators, performing a routing process on the received data from the communication port with reference to the distributed routing table and transferring the data to another routing accelerator. Communication processing method characterized in that it comprises and. 3. A host management unit, a plurality of routing accelerators each transferring packets between the host management unit and another routing accelerator, and each of the routing accelerators transmits packets to and from a network. A communication processing method in a device including a communication port for transmitting and receiving a packet, wherein the host management unit generates / changes relay route information for packet relay, and transfers the relay route information to the plurality of routing accelerators. A communication processing method comprising: distributing, and in each of the routing accelerators, performing a routing process on a received packet from the communication port based on the distributed relay path information, and transferring the packet to another routing accelerator. 4. A processing apparatus comprising: a first processing unit; and a plurality of second processing units, wherein each of the second processing units performs a routing process on a packet received from a network based on relay route information, What is claimed is: 1. A communication processing method in an apparatus configured to transfer a packet to a first processing unit or another second processing unit, wherein the first processing unit generates / changes the relay path information for packet relay. A communication processing method comprising: distributing the relay route information to the plurality of second processing units; and storing the distributed relay route information in each of the second processing units. 5. A processing apparatus comprising: a first processing unit; and a plurality of second processing units, each of the second processing units performing a routing process on a packet received from the network with reference to a routing table, A communication processing method in an apparatus configured to transfer the routing table to a first processing unit or another second processing unit, wherein the first processing unit creates the routing table; A communication processing method, comprising: distributing to a plurality of second processing units; and, in each of the second processing units, storing the distributed routing table. 6. A processing apparatus comprising: a first processing unit; and a plurality of second processing units, wherein each of the second processing units performs a routing process on a packet received from a network based on relay route information, What is claimed is: 1. A communication processing method in an apparatus configured to transfer information to a first processing unit or another second processing unit, wherein the first processing unit exchanges information about a network with another internetwork apparatus. And relaying the relay path information to the plurality of second processing units using a protocol for performing the processing, and in each of the second processing units, the distributed relay path information. A communication processing method characterized by storing 7. A processing system comprising: a first processing unit; and a plurality of second processing units, each of which performs a routing process on a packet received from a network based on relay route information, What is claimed is: 1. A communication processing method in an apparatus configured to transfer the information to a first processing unit or another second processing unit, wherein the first processing unit collects the relay route information from information set by a user. A communication processing method comprising: distributing the relay route information to the plurality of second processing units; and storing the distributed relay route information in each of the second processing units.