JP2002368793A - Inter-network relay device and packet flow identification method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an inter-network relay device that efficiently relays a packet 3 received from a network to other networks. SOLUTION: The inter-network relaying device is provided with a plurality of contents addressable memories (CAMs) 18, 19 to which an entry number is set by each combination of items by using a plurality of items denoting the attribute included in each packet 3 for a retrieval key, a processing retrieval table 20 that stores relay processing information with respect to a packet by each combination of entry numbers obtained by each CAM, a retrieval means 17 that extracts each item of a received packet, retrieves each CAM according to each items to obtain each entry number, combines the entry numbers to generate a retrieval address, and retrieves the processing retrieval table by using the retrieval address to obtain relay processing information corresponding to the packet, and relay means 9, 12 that relay the packet on the basis of the obtained relay processing information.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an inter-network relay device for relaying Internet packets transmitted and received between a plurality of networks, and a packet flow identification method for identifying handling of a packet in the inter-network relay device. .

[0002]

2. Description of the Related Art As shown in FIG.
When information is transmitted and received between the terminals 2a, 2b and 2c belonging to a, 1b and 1c using the Internet, the information is incorporated in a plurality of packets 3 and transmitted and received. When the packet 3 is transmitted and received over a plurality of networks, the transmission and reception of the packet is performed by each of the networks 1a, 1b,
This is performed using the inter-network relay device 5 connected to the communication network 1c via the lines 4a, 4b, and 4c.

FIG. 8 shows an Internet protocol IP.
FIG. 3 is a configuration diagram of a packet 3 conforming to v4 (Internet Protocol version 4). The packet 3 includes a header 3a and information (data) 3b. In the header 3a, a version (Version) of the IP protocol and a header length IHL (I
Internet Header Length), service type TOS (T
ype of Service), packet length (Total Length), identifier ID, flag, fragment offset, lifetime T
Items such as TL (Time to Live), protocol (Protocol), source address SA (Source Address), and destination (destination) address DA (Destination Address) are incorporated.

[0004] The service type TOS indicates the quality of service requested by the packet 3, specifies the priority, requests a route with a small delay, selects a route with a high throughput, and requests a route with a high reliability. Etc. are included.

[0005] Each of the networks 1a to 1c transmits
The inter-network relay device 5 to which the packet 3 having the configuration shown in FIG. 8 is input via the terminal 4c determines the relay destination of the input packet 3, and transfers the packet 3 to the lines 4a to 4c.
4c has a basic function of transmitting to the relay destination networks 1a to 1c.

Further, the inter-network relay device 5 has a function of relaying the packet 3 and, in addition to the original relay function, realizing services such as priority control and bandwidth guarantee in the networks 1a to 1c. Class information indicating various attributes of the packet, such as the service requested by the packet 3, the importance of each packet 3, and the packet type of each packet 3, is determined, and relay processing is performed according to the class information. .

Therefore, the inter-network relay device 5
Needs to have a function of determining the relay destination of the input packet 3 and a function of determining the class information.

FIG. 9 is a block diagram showing a schematic configuration of an inter-network relay device 5 having a relay destination determining function and a class information determining function.

The inter-network relay device 5 includes a physical layer processing unit 6, a data link layer processing unit 7, a packet buffer 8, a packet buffer control unit 9, a network layer processing unit 10, a relay destination / class information table 11, a priority control, It comprises a processing unit 12 and a relay processing control unit 13.

The operation of the inter-network relay device 5 configured as described above will be sequentially described along the flow of processing of the packet 3. The physical layer processing unit 6 applies O to the packet 3 input from the lines 4a to 4c,
The physical layer processing on the SI reference model is performed and transferred to the data link layer processing unit 7. Data link layer processing unit 7
Performs data link layer processing, such as control of a logical line set on a physical line between nodes, on the packet 3 passed through the physical layer processing unit 6, and performs packet buffer control unit 9 in the packet buffer 8. Write to the designated address and report the storage of the packet 3 to the packet buffer 8 to the packet buffer controller 9.

The packet buffer 8 stores the packets 3 processed by the physical layer processing unit 6 and the data link layer processing unit 7.
Is temporarily stored.

The packet buffer controller 9 manages the buffer of the packet buffer 8. That is, the buffer address in the packet buffer 8 where the packet 3 to be transmitted / received next is stored is notified to the data link layer processing unit 7, and the packet 3 to be processed next is notified to the network layer processing unit 10. Notifies the stored buffer address.

In the relay destination / class information table 11,
For each packet 3, the relay destination and class information of the packet 3 are stored for each search address given to the packet 3. Specifically, a logical output port to which the lines 4a to 4c to the networks 1a to 1c that transmit (relay) the packet 3 are stored as the relay destination.

The network layer processing unit 10 obtains a search address based on data of each item such as TOS, SA, DA, etc. designated in advance in the header 3a shown in FIG. 8 of the packet 3 to be processed. Then, the relay destination / class information table 11 is searched with this search address to obtain the relay destination and class information of the packet 3 corresponding to the search address.

Here, the relay destination / class information table 11
A method for generating a search address for searching for relay destination and class information will be described. For example, if the information to be incorporated in the 16-bit search address exceeds 16 bits, a hash operation is used to compress the number of bits of the information exceeding 16 bits to 16 bits or less. Specifically, information to be incorporated into the search address is represented by the service type TOS, the protocol, and the header 3a of the packet 3 shown in FIG.
When SA and DA are set to 5 items, the total of these 80bi
t (= 8bit (TOS) + 8bit (protocol) + 32bit
It is necessary to compress (SA) +32 bits (DA)) into 16 bits using a hash operation.

Further, the network layer processing unit 10 confirms the validity of the packet 3 to be processed by calculating the lifetime TTL, counting the packet length, and the like.

When it is confirmed that the packet 3 is a valid packet 3, the network layer processing unit 10 notifies the packet buffer control unit 9 of the obtained relay destination together with the buffer address, The priority control processing unit 12 is notified of the class information. vice versa,
If the packet 3 is invalid, the packet buffer controller 9
Error information together with the buffer address.

The priority control processing unit 12 performs priority control such as prioritization with other packets 3 waiting to be transmitted based on the class information notified from the network layer processing unit 10, and performs priority control for each packet 3. Determine the transmission timing,
Notify the packet buffer control unit 9.

According to the transmission timing of the packet 3 notified from the priority control processing unit 12, the packet buffer control unit 9 instructs the data link layer processing unit 7 to specify the logical output port of the relay destination together with the buffer address, Notification of transmission.

The data link layer processing unit 7 reads out the packet 3 from the packet buffer 8 in accordance with the instruction of the packet buffer control unit 9 and passes through the physical layer processing unit 6 to the lines 4a to 4c connected to the designated logical output ports. To the packet 3.

A relay processing control unit 13 composed of, for example, a CPU controls the entire inter-network relay processing apparatus 5, initializes various parameters in the processing units 6 to 12, and sets a network layer processing unit. In step 10, processing is performed on the packet 3 for which the relay destination and class information cannot be obtained and processed.

FIG. 10 shows another inter-network relay device 5.
FIG. 2 is a block diagram showing a schematic configuration of a. The same parts as those of the inter-network relay device 5 shown in FIG. 9 are denoted by the same reference numerals, and detailed description of the overlapping parts will be omitted.

The inter-network relay device 5a includes a physical layer processing unit 6, a data link layer processing unit 7, a packet buffer 8, a packet buffer control unit 9, a network layer processing unit 14, a relay destination / class determination CAM (content addressable・ Memory: ContentsAddressable Memory) 1
5, a relay destination / class information table 16 as a processing search table, a priority control processing unit 12, and a relay processing control unit 13.

The different points from the inter-network relay device 5 shown in FIG.
Only the class information table 16 and the network layer processing unit 14 will be described in detail.

Relay destination / class determination CAM (content
Addressable Memory: Contents Addressable Memor
y) 15 is a kind of search table, and this relay destination /
As shown in FIG. 11A, the class determination CAM 15 has a total of 5 logical input ports of the input packet 3, the service type TOS of the corresponding packet 3, the protocol, the source address SA, and the source address DA. One entry number (address) for each item (search key)
Is assigned.

Therefore, for example, each network 1a
To the other networks 1a to 1c from the terminals 2a to 2c of the
Unless the packet 3 is transmitted a plurality of times to the terminals 2a to 2c of c, different entry numbers (addresses) should be assigned to each packet 3. However, among the five items (search keys) from the above-described logical input port to the source address DA, items that are not adopted in the search indicated by “*” are set. Therefore, the item indicated by “*” may be any value. In other words, out of the five items from the logical input port to the destination address DA, each packet 3 that differs only in the item indicated by “*”
Are the same entry number (address).

For example, in entry # 1, TO
S, protocol, and SA are masked (not searched), and only the logical input port and DA are searched. Similarly, in entry # 2, all items of logical input port, TOS, protocol, SA, and DA are to be searched. Further, in entry # 3, only the logical input port, protocol, SA, and DA are searched. Further, in the entry # 4, the logical input port, TOS, and DA are targeted.

Thus, for each entry number (#),
Any number of items (search keys) can be set. That is, an item (search key) to be searched may be selected and set so that a plurality of packets 3 capable of determining the same relay destination and class information have the same entry number (address).

In the relay destination / class information table 16,
As shown in FIG. 2B, for each search address equal to the entry number, the relay destination and class information of the packet 3 belonging to the entry number corresponding to the search address are stored. Specifically, a logical output port to which the lines 4a to 4c to the networks 1a to 1c that transmit (relay) the packet 3 are stored as the relay destination.

The network layer processing unit 14 reads the packet 3 having the designated buffer address from the packet buffer control unit 9 in the packet buffer 8, and reads the packet 3 to be processed in the header 3a shown in FIG. Logical input port, service type TO of the corresponding packet 3
A search is made in the relay / class determination CAM 15 with a total of five items (search keys) of S, protocol, source address SA, and destination address DA to obtain a corresponding entry number (address).

The network layer processing unit 14 searches the relay destination / class information table 16 with the search address using the entry number (address) obtained by the relay destination / class determination CAM 15 as a search address. And the relay destination and class information of the packet 3 corresponding to.

Further, the network layer processing unit 14 confirms the validity of the packet 3 to be processed by calculating the lifetime TTL, counting the packet length, and the like.

When it is confirmed that the packet is a valid packet 3, the network layer processing unit 14 notifies the packet buffer control unit 9 of the obtained relay destination together with the buffer address, 3 is notified to the priority control processing unit 12.
Conversely, in the case of an invalid packet 3, the packet buffer control unit 9 is notified of error information together with the buffer address.

As described above, in the inter-network relay device 5a shown in FIG. 10, the relay destination / class determination CAM
By setting an item indicated by “*” that does not need to be searched for in 15, the number of entry numbers (addresses) set in the relay destination / class determination CAM 15 and the set number of search addresses in the relay destination / class information table 16 are reduced. Can be significantly reduced.

[0035]

However, FIG.
The conventional inter-network relay devices 5, 5 shown in FIG.
In the case of a, the following problems still need to be solved.

That is, in the inter-network relay device 5 shown in FIG. 9, the relay destination / class information table 11 stores the header 3
It is necessary to set a search address that includes information on all of the above-mentioned five items that constitute a. That is, since it is necessary to provide a search address almost every packet 3, the number of search addresses becomes enormous, and the relay destination / class information table 11
Required storage capacity increases.

Further, when the network configuration is changed or a service is added, the number of entries to be registered / deleted from the relay destination / class information table 11 becomes enormous. There is a problem that the operation management cost becomes very large when it is desired to set the quality finely.

Further, since the number of bits that can be handled by the hash search circuit is generally limited, if the bit length of the header 3a of the packet 3 to be subjected to bit number compression is long, it is necessary to repeat the hash operation a plurality of times. However, there is a problem that the calculation time becomes longer and the processing performance deteriorates. Furthermore, since the relay destination / class information table 11 in which the relay destination and class information are stored is referred to using the data obtained by compressing the data by the hash operation as a search address, if the compression ratio is high, the collision frequently occurs and the processing performance is increased. May decrease.

Next, in the inter-network relay apparatus 5a shown in FIG. 10, one entry number is assigned to information including the relay destination determination and the class determination in the relay destination / class determination CAM 15. Therefore, a common entry number (address) is defined for a combination of all target items, and a relay destination is determined based on the entry number (address), although there is an item that is not to be searched depending on the entry number. The search address, the relay destination, and the class information are prepared in the / class information table 16.

As a result, the relay destination / class determination CAM 15
In addition, there is a problem that the required storage capacity of the relay destination / class information table 16 increases. Therefore, there is a concern that the manufacturing cost of the entire inter-network relay device 5a will increase.

The present invention has been made in view of such circumstances, and by adopting a different CAM for each purpose, it is possible to easily cope with a change in specifications and services required by each network, and to implement a CAM. It is possible to reduce the required storage capacity of the entire search table including the network, reduce the manufacturing cost of the entire apparatus, and further improve the relay processing efficiency for packets. An object of the present invention is to provide a method for identifying a packet flow.

[0042]

The present invention is applicable to an inter-network relay device which can be connected to a plurality of networks and relays each packet input from each network to another network.

In order to solve the above problem, in the inter-network relay device of the present invention, a plurality of items indicating the attribute of the packet included in each packet are used as search keys, and an entry number is assigned to each combination of items. Multiple CAMs (content addressable memory) set with
And a processing search table for storing relay processing information for a packet for each combination of entry numbers obtained in each CAM, extracting each item of an input packet, searching each CAM with this item, and searching for each entry Search means for obtaining a search number by combining the entry numbers, generating a search address, searching the processing search table with the search address to obtain the relay processing information corresponding to the packet, and the obtained relay processing. Relay means for relaying the packet based on the information.

In the inter-network relay apparatus configured as described above, a plurality of CAMs are employed according to the type and purpose of the relay processing information necessary for relaying a packet. The number can be reduced compared to the number of combinations when the whole is formed by one CAM. In addition, CAs that search by type or purpose once
Since only the M search is required, the speed of the packet relay process can be increased.

According to another aspect of the invention, the search means in the inter-network relay device of the above-described invention generates a search address by compressing data obtained by combining each entry number by using a hash calculation method. .

In the inter-network relay device configured as described above, since the search address is compressed by performing a hash operation, it is not always necessary to secure an area of an unnecessary combination of entry numbers. Table capacity can be reduced.

According to another aspect of the present invention, the process search table in the inter-network relay device according to the above-described invention stores the relay process information for each combination of the entry numbers before the hash operation and generates the result of the hash operation. A next candidate pointer for sequentially searching for a plurality of pieces of relay processing information having the same search address, wherein the search means searches the processing search table with the search address generated by the hash operation, and finds the searched relay processing information. If the combination of the entry numbers corresponding to... Does not match the combination of the entry numbers of the corresponding search address before the hash operation, the relay processing information specified by the next candidate pointer is searched.

In the inter-network relay device configured as described above, as a result of searching the processing search table with the search address compressed by the hash operation, correct relay processing information can be obtained even if a collision occurs.

In another aspect of the present invention, a plurality of CAMs in the inter-network relay device according to the above-described invention include a relay destination determination CAM in which an entry number designating the same relay destination is set for each combination of items, And a class determination CAM in which an entry number for specifying the same class information indicating the type of packet is set for each combination of. Further, the relay processing information stored in the processing search table includes a relay destination and class information.

The present invention provides a packet flow identification method which is connectable to a plurality of networks, and identifies packet handling in an inter-network relay device which relays each packet input from each network to another network. Applied to

In order to solve the above problem, in the packet flow identification method of the present invention, each item of the input packet is extracted, and an entry number is set for each item in each of these items. A plurality of CAMs (content addressable memories) are searched to obtain respective entry numbers, and the obtained entry numbers are combined to generate a search address. With this search address, a packet is generated for each entry number combination. Then, a process search table storing the relay process information for the packet is searched to obtain the relay process information corresponding to the packet.

Further, another invention is such that, in the above-described packet flow identification method, a combination of entry numbers obtained by searching each CAM is subjected to data compression using a hash calculation method to generate a search address. I have to.

Further, another invention is the packet flow identification method according to the above invention, wherein the relay processing information is stored in the processing search table for each combination of each entry number before the hash calculation, and the hash calculation is performed. As a result of adding a next candidate pointer for sequentially searching a plurality of pieces of relay processing information having the same search address generated as a result and searching the processing search table with the search address generated by the hash operation, the searched relay processing information When the combination of the corresponding entry numbers does not match the combination of the entry numbers of the corresponding search address before the hash operation, the relay processing information specified by the next candidate pointer is searched.

[0054]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of an inter-network relay device 5b employing a packet flow identification method according to an embodiment of the present invention. 9 and 1
The same reference numerals are assigned to the same parts as those of the conventional inter-network relay devices 5, 5a shown in FIG. Therefore, the detailed description of the overlapping part will be omitted.

Further, the inter-network relay device 5b
And the packets 3 input / output to / from the inter-network relay device 5b
Are as shown in FIGS. 7 and 8, respectively.

The inter-network relay device 5 of this embodiment
b denotes a physical layer processing unit 6, a data link layer processing unit 7, a packet buffer 8, a packet buffer control unit 9, a network layer processing unit 17, a relay destination determination CAM 18, a class determination CAM 19, and a relay destination / class as a process search table. It comprises an information table 20, a priority control processing unit 12, and a relay processing control unit 13.

The difference from the conventional inter-network relay devices 5 and 5a shown in FIGS. 9 and 10 is that the relay destination determination CAM
18, the class determination CAM 19, the relay destination / class information table 20, and the network layer processing unit 17 only.
These will be described in detail.

Relay destination determination CAM (Contents Addressable
Memory) 18 is the relay destination / class determination C described above.
As in the case of the AM 15, this is a kind of search table. In the relay destination determination CAM 18, as shown in FIG.
One entry number (address) is assigned to a total of three items (search keys) of the service type TOS and the transmission destination (destination) address DA in the header 3a.

That is, in the relay destination determination CAM 18, one entry number (address) is allocated only to three items (search keys) considered to be necessary for determining the relay destination of the packet 3. Since there is no need to refer to other items in the header 3a of the packet 3 from the beginning, the number of entry numbers (addresses) to be set can be greatly reduced as compared with the number of packets 3.

Further, among the three items (search keys) of the above-described logical input port, service type TOS, and transmission destination address DA, items not employed in the search indicated by “*” are set. Therefore, the item indicated by “*” may be any value. For example, entry #
In No. 1, only the logical input port and DA are to be searched. Similarly, in entry # 2, all items of the logical input port, TOS, and DA are to be searched. Further, in entry # 3, only the logical input port and TOS are to be searched. Further, in the entry # 4, only the logical input port is targeted.

As described above, for each entry number (#),
Any number of items (search keys) can be set according to the relay destination. That is, the search target item (search key) may be selected and set so that the plurality of packets 3 that can be determined to be the same relay destination have the same entry number (address). Therefore, the number of entry numbers (addresses) can be further reduced.

Further, a class determination CAM (Contents Addre
2B, the logical input port of the input packet 3, the service type TOS, the protocol, the transmission source address SA, and the transmission destination in the header 3a of the packet 3, as shown in FIG. One entry number (address) is assigned to a total of five items (search keys) of (destination) address DA.

That is, in the class determination CAM 19, one entry number (address) is assigned only to five items (search keys) considered to be necessary for determining the class information of the packet 3 concerned. ing.

Further, the above-mentioned logical input port, service type TOS, protocol, source address SA,
Of the five items (search key) of the transmission destination address DA,
Items that are not used in the search indicated by “*” are set.
Therefore, the item indicated by “*” may be any value. For example, in the entry # 1, only the source address SA and the destination address DA are searched. Similarly, in entry # 2, all five items of the logical input port, the service type TOS, the protocol, the source address SA, and the destination address DA are to be searched. Further, in entry # 3, the logical input port, protocol, source address S
A, four items of the transmission destination address DA are searched. Further, in the entry # 4, only the logical input port and the service type TOS are targeted.

As described above, for each entry number (#),
Any number of items (search keys) according to the class information can be set. That is, an item (search key) to be searched may be selected and set so that a plurality of packets 3 that can be determined to have the same class information have the same entry number (address). Therefore, the number of entry numbers (addresses) can be further reduced.

In the relay destination / class information table 20,
As shown in FIG. 3, a logical output port and a class information as a relay destination are set for a combination of the relay destination index and the class information index. Further, for the combination of the relay destination index and the class information index, a search address generated by performing a hash operation on the combination of the relay destination index and the class information index to compress the number of bits (data length) is assigned. Have been.

Here, the relay destination index and the class information index are the entry numbers (addresses) of the relay destination determination CAM 18 and the class determination CAM 19, respectively.

Therefore, there are a plurality of combinations of the relay destination index and the class information index having different values for the same search address.

Further, when a collision occurs when a search is performed using a search address employing hash processing, a next point for designating a combination of a next relay destination index having the same search address and a class information index is set.

Furthermore, the presence / absence of transfer of the packet 3 to the relay processing control unit (CPU) 13 and the presence / absence of discard of the packet 3 are set. For example, a packet addressed to the relay device is not relayed and the relay processing control unit (CPU) 1
Take it into 3. Also, for example, the packet 3 that has been determined in advance as being harmful, such as a virus, is discarded.

The network layer processing unit 17 acquires the relay destination and class information of the input packet 3 from the relay destination determination CAM 18, the class determination CAM 19, and the relay destination / class information table 20, and checks the validity of the packet 3 Has the function of determining

As shown in FIG. 4, in the network layer processing section 17, a packet processing control section 21, a packet receiving section 22, a packet storage memory 23, a packet analyzing section 24, a relay processing flow identifying section 25, A packet transmission unit 26 and a priority control processing unit interface 27 are provided.

Next, the internal operation of the network layer processing unit 17 configured as described above will be described step by step. The packet processing control unit 21 sends the packet processing control unit 21
When receiving the reception notification of the packet 3 together with the address of the packet buffer 8 in which the packet 3 is stored, it instructs the packet receiving unit 22 to start the process of reading the packet 3 from the packet buffer 7.

The packet receiving section 22 includes a packet buffer 8
, And stores the packet 3 in the packet storage memory 23. When the storage of the packet 3 is completed, the packet receiving unit 22 notifies the end of the reading process to the packet processing control unit 2.
Notify 1 Upon receiving the read processing end notification from the packet receiving unit 22, the packet control unit 21 notifies the packet analysis unit 24 of the start of the analysis processing.

The packet analysis unit 24 refers to the packet 3 stored in the packet storage memory 23, and
The layer 2 frame format on the SI reference model is specified, and the head position of the layer 3 header is specified. The packet analysis unit 24 stores the analysis result in the packet storage memory 23 and notifies the packet processing control unit 21 of the end of the analysis processing. The reason why the packet analysis unit 24 is provided is that the inter-network relay device 5
This is because it is assumed that there are a plurality of formats in the layer 2 frame format of the packet 3 input to b.

The packet processing control unit 21 instructs the relay processing / flow identification unit 25 to start processing for the packet 3. The relay processing / flow identification unit 25 reads the packet 3 from the packet storage memory 23, confirms the validity of the packet 3, and acquires relay destination and class information. The acquisition of the relay destination and class information is performed using the relay destination determination CAM 18, the class determination CAM 19, and the relay destination / class information table 20 described above.

The relay processing / flow identification unit 25 stores the processing result (validity of the corresponding packet 3, relay destination, class information) in the packet storage memory 23,
1 is notified of the end of the process.

The packet processing controller 21 notifies the packet transmitter 26 of the start of the process of writing the packet 3 to the packet buffer 8. The packet transmitting unit 26 reads out the packet 3 from the packet storage memory 23 and stores it in the packet buffer 8. The packet transmitting unit 26 receives the packet 3
Is completed, the packet processing control unit 21 is notified of the end of the writing process.

The reason why it is necessary to rewrite the packet 3 into the packet buffer 8 is that each item of the header 3a is replaced with another value, or that the header 3a is inserted by the encapsulation and decapsulation of the header 3a.・ This is because deletion may be required.

The packet processing control section 21 notifies the priority control processing section interface 27 of the start of processing.
The priority control processing unit interface 27 notifies the priority control processing unit 12 of the class information stored in the packet storage memory 23 together with the address of the packet buffer 8 in which the corresponding packet 3 is stored.

When the end of the processing is notified from the priority control processing section interface 27, the packet processing control section 21
The packet buffer control unit 9 is notified of the end of the network layer processing, together with the address of the packet buffer 8 in which the target packet 3 is stored and the relay destination.

Next, using the relay destination determination CAM 18, the class determination CAM 19, and the relay destination / class information table 20, a relay process for obtaining the relay destination and class information of the packet 3
The detailed configuration and operation of the flow identification unit 25 will be described.

In the relay processing / flow identification unit 25,
As shown in FIG. 5, the packet storage memory control unit 28, the header processing unit 29, the relay destination determination CAM control unit 30, the class determination CAM control unit 31, the address generation unit 32, the relay destination /
A class information table control unit 33 is provided.

The specific processing operation of the relay processing / flow identification unit 25 is described in the relay destination determination CAM 18 and the class determination CAM 19 in FIG. 2 and the relay destination / class information table 2 in FIG.
0, and a flowchart shown in FIG.

First, upon receiving a notification of the start of processing from the packet processing control unit 21 in FIG. 4, the packet storage memory control unit 28 in FIG. 5 receives the notification from the packet analysis unit 24 stored in the packet storage memory 23. Referring to the head position of the header 3a of the packet 3, the packet 3
Is read. The packet data composed of each item data constituting the header 3a of the obtained packet 3 is sent to the header processing unit 29, the relay destination determination CAM control unit 30, and the class determination CA
It is passed to the M control unit 31.

The header processing unit 29 confirms the validity of the packet 3. The validity check of the packet 3 includes checking that the version (Version) in the header 3a of the packet 3 shown in FIG.
ernet Header Length) is more than "5", packet length (Total Length) is correct,
The survival time TTL is subtracted, the survival time TTL after the subtraction is confirmed to be “1” or more, and a checksum (Header C)
hecksum) is correct.

After confirming the validity of the packet 3, the header processing unit 29 notifies the packet storage memory control unit 28 of the confirmation result (whether the packet is valid or not).

The relay destination determination CAM control unit 30 determines the TOS, the destination address DA, and the corresponding packet 3 of each item data of the given packet data of the corresponding packet 3.
Using the logical input port of (1), the relay destination determination CAM 18 shown in FIG. 2A is searched to determine whether there is an entry number (address) in which the value of each item matches. The relay destination determination CAM control unit 30 notifies the address generation unit 32 of the matching entry number (address). When there is no matching entry number (address), the relay destination determination CAM control unit 30 notifies the address generation unit 32 that there is no matching entry number (address) (step S1 in FIG. 6).

Similarly, the class determination CAM control unit 31
TOS, protocol, source address S in each item data of the packet data of the given packet 3
A, the destination address DA, and the logical input port of the corresponding packet 3, the class determination CAM1 shown in FIG.
9 to determine whether there is an entry number (address) that matches the value of each item. Class determination CAM control unit 31
Notifies the address generation unit 32 of the matching entry number (address). The class determination CAM control unit 31
Notifies the address generation unit 32 that no matching entry number (address) exists (step S2 in FIG. 6).

The address generation unit 32 is a relay destination determination CAM
Based on the address of each entry number notified from the control unit 30 and the class determination CAM control unit 31, a search address for searching the relay destination / class information table 20 is generated.

In this case, the relay destination / class information table 2
Data compression of the search address is performed to reduce the required storage capacity of zero. That is, the entry numbers notified from the relay destination determination CAM control unit 30 and the class determination CAM control unit 31 are called a relay destination index and a class information index, respectively.

Then, the address generation unit 32 performs a hash operation on a composite of the relay destination index and the class information index to compress the number of bits (data length) as a search address, and uses the created search address as the relay destination. /
It notifies the class information table control unit 33 (step S3 in FIG. 6).

The relay destination / class information table control unit 33 refers to the relay destination / class information table 20 using the search address notified from the address generation unit 28 (step S4 in FIG. 6).

In this case, since the search address space is compressed by the hash operation, there is a possibility that a collision for searching for another combination of the relay destination index and the class information index having the same search address may occur.

Therefore, the relay destination / class information table 2
The relay destination index and the class information index designated by the search address 0 are compared with the entry numbers (addresses) of the corresponding search address before compression, and it is determined whether or not the referenced search address should be used ( Step S5 in FIG. 6).

If each index specified by the search address matches each entry number (address) before compression, the relay destination (logical output port) specified by the search address in the relay destination / class information table 20 and The class information is acquired (Step S6 in FIG. 6).

If the indexes do not match, the next pointer specified by the search address is obtained (step S7 in FIG. 6), and the relay destination index and the class information in the relay destination / class information table 20 specified by the next pointer are obtained. Look up the index again. Thereafter, until the index matches or the next pointer becomes a null pointer (FIG. 6).
Step S8) Similar processing is repeated.

Note that the null pointer is a special pointer indicating that there is no subsequent relay destination index and class information index, and a relay destination (logical output port) corresponding to the relay destination / class information table 20.
And that the class information is not registered.

As shown in the flowchart of FIG. 6, the relay destination determination CA
If there is no search address whose index matches in the search of the M18 or the search of the class determination CAM 19, or if a null pointer is finally obtained, the packet is processed by the relay processing control unit (CPU) 13 is instructed to transfer (step S9 in FIG. 6).

If the respective indexes match,
The relay destination / class information table control unit 33 notifies the packet storage memory control unit 28 of the obtained relay destination (logical output port), class information, and the like.

The packet storage memory control unit 28 writes the packet 3 into the packet storage memory 23 together with the obtained relay destination (logical output port), class information, and the like.
After the writing to the packet storage memory 23 is completed, the packet storage memory control unit 28 notifies the packet processing control unit 21 of the end of the processing.

The overall operation of the inter-network relay device 5b incorporating the network layer processing unit 17 for performing the processing operation on the input packet 3 will be described with reference to FIG.

The physical layer processing unit 6 applies an OS such as control of a physical line to the packet 3 input from the lines 4a to 4c.
It performs the processing of the physical layer on the I reference model and transfers it to the data link layer processing unit 7. Data link layer processing unit 7
Performs data link layer processing, such as control of a logical line set on a physical line between nodes, on the packet 3 passed through the physical layer processing unit 6, and performs packet buffer control unit 9 in the packet buffer 8. Write to the designated address and report the storage of the packet 3 to the packet buffer 8 to the packet buffer controller 9.

The packet buffer 8 stores the packet 3 processed by the physical layer processing unit 6 and the data link layer processing unit 7.
Is temporarily stored. The packet buffer control unit 9 manages the buffer of the packet buffer 8. That is, the packet 3 to be transmitted / received next to the data link layer processing unit 7
Is notified of the buffer address in the packet buffer 8 where the packet 3 is stored, and the buffer address where the packet 3 to be processed next is stored is notified to the network layer processing unit 10.

The network layer processing unit 17 obtains the relay destination (logical output port) and class information of the packet 3 input from the relay destination determination CAM 18, the class determination CAM 19, and the relay destination / class information table 20, and 3 is determined. Then, when it is confirmed that the packet is a valid packet 3, the network layer processing unit 17 notifies the packet buffer control unit 9 of the obtained relay destination (logical output port) together with the buffer address, and It notifies the priority control processing unit 12 of the class information of the packet. Conversely, if the packet 3 is invalid, the packet buffer control unit 9 is notified of the error information together with the buffer address.

The priority control processing unit 12 performs priority control such as prioritization with other packets 3 waiting to be transmitted based on the class information notified from the network layer processing unit 17, and performs priority control for each packet 3. Determine the transmission timing,
Notify the packet buffer control unit 9.

According to the transmission timing of the packet 3 notified from the priority control processing unit 12, the packet buffer control unit 9 designates the logical output port of the relay destination together with the buffer address to the data link layer processing unit 7, Notification of transmission.

The data link layer processing unit 7 reads out the packet 3 from the packet buffer 8 in accordance with the instruction of the packet buffer control unit 9, and passes through the physical layer processing unit 6 to the lines 4 a to 4 c connected to the designated logical output ports. To the packet 3.

The relay processing control section 13 controls the entire inter-network relay processing apparatus 5b, initializes various parameters in each processing section, and sets a relay destination (logical output port) in the network layer processing section 17. And a process for the packet 3 for which the class information cannot be obtained and processed.

In the inter-network relay apparatus 5b thus configured, a relay destination determination CAM 18 for determining the relay destination of the input packet 3 and a class for determining the class information of the input packet 3 Judgment CA
M19 and M19 are realized by separate CAMs.

Therefore, in the relay destination determination CAM 18, only the items necessary for determining the relay destination among the items set in the logical input port of the packet 3 and the header 3a can be adopted as the search items. Similarly, class determination CA
In M19, of the items set in the logical input port of the packet 3 and the header 3a, only items necessary for determining the class information can be adopted as search items.

As a result, the number of entry numbers (addresses) to be set by combining the relay destination determination CAM 18 and the class determination CAM 19 is set by the relay destination / class determination CAM 15 in the conventional inter-network relay device 5a shown in FIG. The number can be greatly reduced as compared with the number of entry numbers (addresses) to be provided.

Incidentally, the relay destination / class determination CAM1 in the conventional inter-network relay device 5a shown in FIG.
In 5, for items related to either the relay destination or the class information, it is necessary to add the relevant item to the search target, and the relay destination / class determination CAM 15 is set as many as the number of combinations of items is. The number of entry numbers (addresses) to be increased increases.

Therefore, the number of search addresses set in the relay destination / class information table 20 can be greatly reduced as compared with the relay destination / class information table 16 in the conventional inter-network relay device 5a shown in FIG.

As described above, the number of entry numbers (addresses) and the number of search addresses in each of the determination CAMs 15 and 16 and the table 20 can be reduced. The search (acquisition) processing efficiency can be greatly improved, the storage capacity required for each of the determination CAMs 15, 16 and the table 20 can be significantly reduced, and the manufacturing cost of the entire inter-network relay device 5b can be reduced.

Further, since the relay destination determination CAM 18 and the class determination CAM 19 are realized by different CAMs, the connection configuration of the networks 1a, 1b, and 1c may change.
When the service or specification to be provided is changed, the storage contents of the relay destination determination CAM 18 and the class determination CAM 19 need to be changed, but only the determination CAMs 18 and 19 related to the relay destination or the class information are changed. Therefore, it is possible to flexibly cope with a change in the specifications of the network system or a change in the service.

Further, the relay destination / class information table 20
A plurality of relay destinations and class information having the same search address generated as a result of the hash operation are stored in each area of the relay destination / class information table 20 while compressing data using a hash operation method for a search address for searching for A next point for searching in order is provided.

Therefore, as a result of searching the relay destination / class information table 20 with the search address compressed by the hash operation, correct relay destination and class information can be obtained even if a collision occurs.

[0119]

As described above, the inter-network relay device and the packet flow identification method of the present invention employ a plurality of CAMs having different combinations of items for each purpose. Therefore, it is possible to easily cope with a change in specifications and services required by each network, to reduce the required storage capacity of the entire search table including the CAM, to reduce the manufacturing cost of the entire device, and to perform a relay process for packets. Efficiency can be improved.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a schematic configuration of an inter-network relay device according to an embodiment of the present invention;

FIG. 2 is a diagram showing storage contents of a relay destination determination CAM and a class determination CAM formed in the inter-network relay device.

FIG. 3 is a view showing the storage contents of a relay destination / class information table formed in the inter-network relay device.

FIG. 4 is a block diagram showing a configuration of a network layer processing unit in the inter-network relay device.

FIG. 5 is a block diagram showing a detailed configuration of a relay processing flow identification unit in the network layer processing unit;

FIG. 6 is a flowchart showing a processing operation of a relay processing flow identification unit in the inter-network relay apparatus.

FIG. 7 is a schematic diagram showing a general network system.

FIG. 8 is a format diagram showing a configuration of a packet transmitted and received between networks.

FIG. 9 is a block diagram showing a schematic configuration of a conventional inter-network relay device.

FIG. 10 is a block diagram showing a schematic configuration of another conventional inter-network relay device.

FIG. 11 is a diagram showing storage contents of a relay destination / class determination CAM and a relay destination / class information table formed in the conventional inter-network relay device.

[Explanation of symbols]

 1a, 1b, 1c: Network 5, 5a, 5b: Inter-network relay device 6: Physical layer processing unit 7: Data link layer processing unit 8: Packet buffer 9: Packet buffer control unit 12: Priority control processing unit 13: Relay processing Control unit (CPU) 17 Network layer processing unit 18 Relay destination determination CAM 19 Class determination CAM 20 Relay destination / class information table 25 Relay processing / flow identification unit

Claims (7)

[Claims] Claims: 1. It is connectable to a plurality of networks,
In an inter-network relay device that relays each packet input from each network to another network, a plurality of items indicating attributes of the packet included in each packet are used as search keys, and an entry number is set for each combination of items. A plurality of set CAMs (content addressable memories); a processing search table for storing relay processing information for the packets for each combination of entry numbers obtained by the CAMs; Extracting, searching each CAM with each item to obtain each entry number, generating a search address by combining each entry number,
Search means for searching the processing search table with the search address to obtain relay processing information corresponding to the packet, and relay means for relaying the packet based on the obtained relay processing information. Inter-network relay device. 2. The inter-network relay device according to claim 1, wherein the search means generates a search address by compressing data obtained by combining the respective entry numbers using a hash calculation method. 3. The process search table stores relay process information for each combination of entry numbers before hash calculation, and sequentially stores a plurality of relay process information having the same search address generated as a result of hash calculation. The search means has a next candidate pointer for searching, and as a result of searching the processing search table with a search address generated by a hash operation, a combination of an entry number corresponding to the searched relay processing information is a search address. 3. The inter-network relay device according to claim 2, wherein when the combination does not match the entry number before the hash operation, the relay processing information specified by the next candidate pointer is searched. 4. The plurality of CAMs include a relay destination determination CAM in which an entry number designating the same relay destination is set for each combination of items, and the same class information indicating a packet type for each combination of items. The relay processing information stored in the processing search table is configured by a class determination CAM in which an entry number to be specified is set, and the relay processing information is configured by a relay destination and class information. 2. The inter-network relay device according to claim 1. 5. Connectable to a plurality of networks,
In a packet flow identification method for identifying handling of the packet in an inter-network relay device that relays each packet input from each network to another network, extracting each item of the input packet, A plurality of CAMs (content addressable memories) in which an entry number is set for each combination of items to obtain each entry number, and generate a search address by combining the obtained entry numbers; A packet flow identification method comprising: searching a processing search table storing relay processing information for the packet for each combination of entry numbers using the search address; and obtaining relay processing information corresponding to the packet. 6. The packet flow identification according to claim 5, wherein a combination of the entry numbers obtained by searching each of said CAMs is subjected to data compression using a hash calculation method to generate a search address. Method. 7. A process for storing a plurality of pieces of relay processing information having the same search address generated as a result of a hash operation, wherein the plurality of pieces of relay processing information having the same search address generated as a result of the hash operation are stored in the processing search table for each combination of each entry number before performing a hash operation. As a result of adding the next candidate pointer for the search in order and searching the processing search table with the search address generated by the hash operation, the combination of the entry numbers corresponding to the searched relay processing information is the hash operation of the relevant search address. If there is a mismatch with the previous entry number combination,
7. The packet flow identification method according to claim 6, wherein the relay processing information specified by the next candidate pointer is searched.