JP2002271398A - Network transmission device and method of transmission - Google Patents

Abstract

PROBLEM TO BE SOLVED: To remove invalid data embedded for matching at high speed in exchanging header information of a packet. SOLUTION: When a new header information appropriate to a frame format is created and exchanged for a receiving packet and mismatching section occurs at a boundary between the new header information to be exchanged and the packet data, the invalid data are embedded and a valid/invalid flag is set to every byte-unit data. In this state, after byte unit data including invalid data in the write signal are stored in a plurality of storing means 211 to 214 for a plurality of bytes, a high speed switching means 22 is controlled based on the valid/invalid flag of the byte-unit data to remove the invalid data by selecting output paths of a plurality of storing means and sort to output a plurality of valid data in a given order.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a network relay device for transferring packets between a plurality of networks and a relay method thereof, and more particularly to a network relay device having a header information replacement function and a relay method thereof.

[0002]

2. Description of the Related Art In a relay device having a function of converting a frame format, an input packet shown in FIG. 8A received by a receiving interface is stored in a storage device, and then the packet is converted according to the frame format. Although the header information is transferred with new header information being transferred, the difference in length between the header information before replacement and the new header information after replacement is different from the transfer unit. In this case, the boundary between the new header information and the data as shown in FIG.

Therefore, after embedding invalid data as shown by hatching in FIG. 1C in the inconsistent portion, the invalid data is removed by a subsequent invalid data removing means, and new header information and new header information are added. Create an output packet by aligning the data and output.

Heretofore, the above-mentioned means for removing invalid data have been realized by means of hardware processing and software processing.

[0005] The former hardware-based invalid data removing means includes:
As shown in FIG. 9, the internal bus circuit 51 is
Four 8-bit FIFOs (First in First ou
t: first input destination output) memory 52₁~ 52_Four, 4 on the output side
8-bit FIFO memory 53 ₁~ 53_FourIs connected
You. Then, based on the write enable signal, the invalid data
The byte data including the data is stored in each FIFO memory 52.₁~
52_FourWhile the byte enable signal is
1 bus via the shared bus 51a of the internal bus 51
Data is transferred in units of byte data, but invalid data
Are removed without being transferred, and only valid data is
FO memory 53₁~ 53_FourWrite to_.Read enable
Upon receiving the signal, the output side FIFO memory 53₁~ 53_FourFrom
An output packet, which is 4-byte data, is output.

On the other hand, the latter software-like invalid data removing means is provided with a data storage buffer 60 in addition to an input buffer 58 and an output buffer 59 on a bus line 57 derived from a processor 56 as shown in FIG. , The processor 56 sequentially reads the packet data stored in the input buffer 57 and accumulates the packet data in the data accumulation buffer 60. When sufficient data is accumulated in the data accumulation buffer 60, each byte unit data (8 bits Data), the invalid data is removed, and the data is written to the output buffer 59.

[0007]

SUMMARY OF THE INVENTION However,
Of such invalid data removing means, the former internal bus 51 is used.
Are provided with each FIFO memory 52₁~ 52_FourFrom 1
Byte unit data can be output, but 1
Output FIFO memory 53 narrowed down to bytes ₁~ 5
3_FourIn recent years, the speed of lines has been increasing.
Despite this, high-speed processing is extremely difficult and lacks practicality.
Problem.

The latter software-based invalid data removing means requires a data storage buffer 60 and invalidates the packet data stored in the input buffer 58 when the data storage buffer 60 has sufficiently accumulated the packet data. Since invalid data is removed while judging the existence of data, and then the write processing is performed on the output buffer 59, the bus width inside the device tends to be multi-bit with the increase of the line speed. It is difficult to remove invalid data at high speed by byte-by-byte transfer, and it is difficult to realize a high-speed request.

The present invention has been made in view of the above circumstances, and a network relay apparatus that executes byte alignment while removing invalid data embedded for consistency at a high speed when replacing header information of a packet, and a network relay apparatus therefor. An object is to provide a relay method.

It is another object of the present invention to provide a network relay device and a relay method for performing a header replacement process without reducing the throughput of packet data transfer.

[0011]

(1) In order to solve the above-mentioned problems, a network relay device according to the present invention comprises:
For a received packet, invalid data is embedded when a boundary between new header information that is replaced according to a frame format and data of the packet is mismatched, and a valid / invalid flag is set for each byte of data. Means, and invalid data removing means for arranging and outputting a plurality of bytes of valid data in a predetermined order while removing the invalid data based on the set flag.

According to the present invention having the above-described structure, when embedding invalid data at the boundary between new header information and packet data, the data of each byte data including this invalid data is included. The valid / invalid flag is set bit by bit, and the invalid data is removed and the byte is sorted in a predetermined order based on this flag. It is possible, and it is possible to sufficiently cope with an increase in the line speed.

(2) The network relay device according to the present invention embeds invalid data in a received packet at a boundary between each new header information to be replaced according to a plurality of different frame formats and the same data of the packet. Means for setting a valid / invalid flag for each byte of data, and a frame format for aligning and outputting a plurality of bytes of valid data in a predetermined order while removing the invalid data based on the set flags. And a corresponding invalid data removing unit.

According to the present invention having the above-described configuration, the same operation and effect as those of the above item (1) can be obtained, and new header information corresponding to different frame formats can be added to the same data of a packet. It is possible to transfer to a plurality of networks in a multicast manner.

[0015] (3) A network relay method according to the present invention includes a header generation step of generating new header information for a received packet in accordance with a frame format;
A data embedding step of setting a valid / invalid flag for each byte of data while embedding invalid data when inconsistency occurs at the boundary between the generated new header information and the data of the packet; Removing the invalid data based on the valid / invalid flag, and aligning and outputting a plurality of bytes of valid data.

When new header information is generated for the same received packet according to a plurality of different frame formats, it can be realized by the same steps as described above.

According to the present invention, by using the above-described method, new header information according to the frame format is generated in the header generation stage. Are different,
An inconsistency occurs between the new header information and the data of the packet, but the invalid data is embedded and transferred with consistency. However, the invalid data is originally invalid and needs to be removed. A 1-bit valid / invalid flag is set for each data so that high-speed determination is possible.

In the byte alignment stage, while checking the valid / invalid flag for each byte of data, the invalid data is removed in the case of the invalid flag, and a plurality of byte data is deleted in the case of valid data. If the sorting process is performed and output in a predetermined order, output packets can be created at a high speed.

[0019]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a functional block diagram showing the inside of a network relay device according to an embodiment of the present invention.

This network relay device separates the received IP packet into header information and data, stores the data in the header storage area and the data storage area of the storage device 11, and stores the data in accordance with the frame format. A header conversion means for converting the header data into new header information for replacement; and invalid data when an inconsistency occurs at the boundary between the new header information and the packet data. Invalid data embedding means 14 that embeds a valid / invalid flag for each byte of data.
And invalid data removing means 15 for removing the invalid data in byte units based on the flag set by the invalid data embedding means 14 and generating an output packet for executing byte alignment on the valid data. ing.

The data separating / storing means 12 stores the header information and the data in the storage device 11 separately. For example, the data separation / storage means 12 may store the information in one storage device 11 in different areas or store the data in separate storage devices. It may be stored.

As shown in FIG. 2, the header conversion means 13 is provided with a header table 16 for storing a plurality of replacement header information 1a, 1b,... In advance in consideration of various frame formats. Means is provided for analyzing an IP header, a protocol, and the like existing in the header information, and searching and generating new header information to be replaced from the header table 16 based on the analysis result.

The invalid data embedding means 14 transfers new header information corresponding to the frame format converted by the header converting means 13 to the data in the data storage area, and transfers the new header information to the new header information to be replaced with the previous header information. If the length is different from the information and the difference is not in the data transfer unit (for example, Word unit), an inconsistent portion occurs between the new header information and the data. On the other hand, it has a function of setting a 1-bit flag of "1" for valid data and "0" for invalid data, for example.

The invalid data removing means 15 will be described later, and will not be described here.

FIG. 3 is a functional block diagram showing the inside of the network relay device according to another embodiment of the present invention.

This network relay device is a network relay device to which a multicast method of applying new header information based on a plurality of different frame formats to the same data of a packet and transferring the same to a plurality of users is applied.

More specifically, this relay device is divided into header information and data of an input packet by the data separating / storing means 12, and is stored in a header storage area and a data storage area of the storage device 11 respectively. It is the same as FIG.

In the apparatus according to the present invention, the subsequent processing including the header conversion means 13 is different.

First, when receiving a multicast instruction from outside, the header conversion means 13 analyzes the header information,
Using the header table 16 described above, a plurality of new header information to be replaced is searched and generated based on the frame format to be transferred, and transmitted to the corresponding invalid data embedding means 14 ₁ , 14 ₂ ,. I do. Each of the invalid data embedding means 14 ₁ , 14 ₂ ,..., 14n
Replaces the same data in the data storage area with new header information according to the frame format converted by the header conversion means 13, but since the length of each header information is different, it is the same as each new header information. When an inconsistent portion occurs between the data and the invalid data, the invalid data is embedded in the inconsistent portion. On the other hand, for valid data, for example, “1”, and for invalid data, for example, “0”. Set a flag.

Therefore, each succeeding invalid data removing means 15
_1, 15 ₂ ..., 15n is the invalid data embedding unit 14 ₁ corresponding, 14 ₂ ..., and generates an output packet by removing the invalid data in bytes based on the flag set by 14n .

Next, the invalid data removing means 15 (15 ₁ ,
15 _2, ..., will be described with reference to the drawings configuration of including 15n).

FIG. 4 is a block diagram showing an embodiment of the invalid data removing means 15.

The invalid data removing means 15 stores four bytes of byte-based data including invalid data output from the invalid data embedding means 14 based on a write enable signal, for example, four FIFO memories 21 _{1 to} 21.
_4, with respect to byte data output from the respective FIFO memories 21 ₁ to 21 _4, invalid data embedding unit 1
Based on the byte enable signal, which is the result of determining the valid / invalid flag for each byte of data set in step 4,
While removing the invalid data while selecting switching the output path of the FIFO memories 21 ₁ to 21 _4, and 4 × 4 switch 22 having a high-speed switching function of transferring byte data of plural bytes young turn, this 4 A predetermined number of valid byte units of data transferred from the × 4 switch 22 are sequentially stored, and four, for example, FIFO memories 23 output under a read enable signal.
It is constituted by a ₁ to 23 _4.

The operation of the invalid data removing means 15 as described above will be described with reference to FIG.

Now, from the invalid data embedding means 14,
When 8-bit byte data including invalid is sent, the FIFO memory 21 receives a write enable signal.
_{If 1} , ₂ , 21 ₃ and 21 ₄ are in the write mode, FI
The FO memories 21 ₁ , 21 ₂ , 21 ₃ , and 21 ₄ are shown in FIG.
As shown in FIG.
2, B3 and B4 are written and output.

At this time, the byte unit data B1, B2,
External based on the judgment of the flag added to B3 and B4
Input the byte enable signal which is the switch control signal
By doing so, the FIFO memory 21₁-23₁, 21_Two
-23_Two, 21_Three-23_Three, 21 _Four-23_FourOutput path
The 4 × 4 switch 22 performs the switching operation so that
Alignment is performed, so in byte units as shown in (e)
Data B1, B2, B3, and B4 operate the 4 × 4 switch 22.
pass.

[0038] Thereafter, on the basis of the read enable signal, is set to read mode FIFO memory 23 _1, 23 _2, 23 _3, 23 _4, 4 bytes left side direction from the shown right end shown in the drawing (f) Byte data for each,
That is, byte unit data B1, B2, B3, and B4 can be output.

Accordingly, by setting the write enable signal and the read enable signal at a predetermined timing, the subsequent FIFO memories 23 ₁ , 23 ₂ , 23 ₃ , 2
3 ₄ When the byte data B1, B2, B3, B4 are stored, the previous stage of the FIFO memory 21 _1, 21 _2, 21
_3, 21 bytes data B5, the ₄ B6, B7, B8
Is stored.

Therefore, the FIFO memories 21 ₁ , 21 ₂ , 2
1 _3, 21 _4-byte data block B5, B6, B7, B8
Is stored, the byte enable signal is input under the setting flag, and the 4 × 4 switch 22 forms the output path of the FIFO memories 21 _{1 to} 23 ₁ and 21 _{2 to} 23 ₂ and the byte unit data B5 , B6, but the next byte unit data B7, B8 are set with an invalid flag, so the FIFO memory 21 in the 4 × 4 switch 22
_3, 21 ₄ by opening and removing the invalid data,
If an output path in the 4 × 4 switch 22 is formed so that the next smaller byte-order data B9 and B10 are inserted into the removed portion, 4-byte data as shown in FIG. it can. That is, it is possible to create packet data in which header information and data are completely aligned while removing invalid data.

Therefore, according to the above-described embodiment, a plurality of storage means for storing data in units of bytes are provided on the input / output side of the high-speed switching means which is the 4 × 4 switch 22 to enable / disable. Based on a switch control signal output based on a flag attached to data, the high-speed switching means forms an output path so that only valid data passes and the byte is aligned, thereby removing invalid data at high speed. In addition, byte alignment becomes possible, and high-speed processing that can sufficiently cope with an increase in line speed becomes possible.

FIG. 6 shows the invalid data removing means 15 (15 ₁ , 15 ₁ ,
15 ₂ (including 15 ₂ ,..., 15 n).

The invalid data removing means 15 takes in 8-bit valid / invalid byte-unit data to which a 1-bit valid / invalid flag added from the invalid data embedding means 14 is added, and adds it to each byte-unit data. The flag is separated into each byte unit data and the flag is output as a byte enable signal which is a flag of 7 byte unit data.
1, B2, B3, B4, next is 3-byte data B5
A data extraction / output means 31 for outputting B6 and B7; a two-stage configuration for storing 8-bit byte-unit data; (4) validity judging means 32 for judging validity / invalidity from the content of the byte enable signal and outputting a selection signal for selecting a set of valid data 4 bytes
And selection means 33 (1) to 33 (4) for selecting and outputting 4-byte data in units of bytes in accordance with the selection signal output from the validity judgment means 32.

The operation of the invalid data removing means 15 will be described with reference to FIG.

When the 8-bit valid / invalid byte unit data to which the valid / invalid flag is added is sent from the invalid data embedding unit 14, the data extraction / output unit 31 outputs the flag added to each byte unit data. And a byte enable signal, which is a 7-byte data flag, is sent to the validity judging means 32. The byte group data is sent to the register groups R1 (1) to R1 (4) at the preceding stage. Is sent.

As a result, at a certain point in time, the register units R2 (1) to R2 (4) at the subsequent stage
2, B3, and B4 are stored in the register groups R1 (1) to R1 (3) at the preceding stage in byte unit data B5,
B6 and B7 are stored, but the validity determination means 3
In FIG. 7, a selection signal for selecting valid 4-byte data in ascending order from a byte enable signal which is a flag of 7-byte data, that is, as shown in FIG. A selection signal for selecting the output of R2 (4) is sent to the selection means 33.

Therefore, the selection means 33 (1) to 33 (4)
Selects and outputs byte unit data B1, B2, B3, and B4, which are the outputs of the register groups R2 (1) to R2 (4) at the subsequent stage.

Subsequently, the register group R2 (1) at the subsequent stage
Byte data B5, B6, B7, B in R2 (4)
8 is stored, the preceding register group R1 (1)
ＲR1 (3) contains byte data B9, B10, B1
1 is stored. At this time, the validity judging means 3
7 is a selection signal for selecting four bytes of valid byte-unit data in ascending order from the byte-unit data B5 from the byte enable signal which is a flag of the 7-byte unit data after the byte-unit data B5, that is, FIG. As shown in (1), a selection signal for selecting the output of the register group R2 (1), R2 (2) of the succeeding stage and the register group R1 (1), R1 (2) of the preceding stage is sent to the selecting means 33.

Here, byte unit data B7, B8
Since the validity judging unit 32 naturally judges invalid data from the corresponding flag and does not select it, the removal processing is performed.

Therefore, according to the above-described embodiment, four bytes of valid byte-unit data are selectively output in ascending order from the valid byte-unit data from the outputs of the two-stage register group. Therefore, there is no need to have a special data storage buffer for the packet data format conversion, and the delay time of data transfer can be minimized.

Further, while effectively utilizing the two-stage register group and the selection means, the effective 4
By selecting byte-by-byte data, it is possible to quickly realize a process of adjusting a mismatch between header information and data without reducing throughput.

In each figure, for convenience of explanation, 4
Although the data transfer of the byte-by-byte data has been described, it is possible to sufficiently cope with the expansion of the bus width. That is, n
In the case of transfer in units of bytes, based on the basic configuration of FIG.
Effective n bytes may be selected in ascending order from n-1 bytes. Unlike the method of determining whether data is valid or invalid and always transferring data in byte units, four bytes are simultaneously output, so that high-speed processing can be realized with relatively easy adoption of the configuration.

Next, the invention of a method using the above-described relay device will be described.

The network relay method according to the present invention comprises:
The header information of the received packet is analyzed for the received packet, and new header information according to the frame format is generated (header generation stage), but the length of the previous header information is different from the length of the new header information of the present time. Therefore, if an inconsistency occurs between the new header information and the data of the packet, the valid / invalid flag is set for each byte of data while embedding invalid data in the inconsistency (data embedding stage).

Thereafter, for the byte data including the invalid data, the valid / invalid flag assigned to each byte data is determined, and the invalid data is removed. Output while performing the process of aligning the unit data (byte alignment stage).

When new header information is generated for the same received packet in accordance with a plurality of different frame formats, it can be realized by the same steps as described above.

Therefore, while checking the valid / invalid flag for each byte of data, the invalid data is removed in the case of the invalid flag, and a predetermined order is set for a plurality of byte data in the case of valid data. Since the sorting process is performed and output as described above, an output packet can be created at a high speed.

In addition, the present invention is not limited to the above-described embodiment, and can be variously modified and implemented without departing from the gist thereof. Further, the embodiments can be implemented in combination as much as possible, and in that case, the effect of the combination can be obtained. Furthermore, each of the above embodiments includes various upper and lower stage inventions, and various inventions can be extracted by appropriately combining a plurality of disclosed components. For example, if an invention is extracted because some constituent elements can be omitted from all the constituent elements described in Means for Solving the Problems, the omitted part may be omitted when implementing the extracted invention. Is appropriately supplemented by well-known conventional techniques.

[0059]

As described above, according to the present invention, it is possible to execute byte alignment while removing invalid data embedded for consistency at the time of replacement of header information of a packet at a high speed. It is an object of the present invention to provide a network relay device and a relay method thereof which can sufficiently cope with the change of the network.

Further, the present invention can provide a network relay apparatus and a relay method capable of removing invalid data without reducing the throughput of packet data transfer.

[Brief description of the drawings]

FIG. 1 is an internal functional configuration diagram showing an embodiment of a network relay device according to the present invention.

FIG. 2 is a diagram illustrating header information to be replaced according to a frame format.

FIG. 3 is an internal functional configuration diagram showing another embodiment of the network relay device according to the present invention.

FIG. 4 is a configuration diagram showing an example of an invalid data removing unit shown in FIGS. 1 and 3;

FIG. 5 is a view for explaining the operation of the invalid data removing means shown in FIG. 4;

FIG. 6 is a configuration diagram showing another example of an invalid data removing unit.

FIG. 7 is a view for explaining the operation of the invalid data removing means shown in FIG. 6;

FIG. 8 is a view for explaining an inconsistency between new header information for replacement and data.

FIG. 9 is a configuration diagram of a conventional invalid data removing unit.

FIG. 10 is a configuration diagram of another conventional invalid data removing unit.

[Explanation of symbols]

11 ... storage device 12 ... data separating storage unit 13 ... header converting means 14 ... invalid data embedding unit 15 ... invalid data removing unit 21 ₁ through 21 ₄ ... FIFO memory 22 ... 4 × 4 switches 23 ₁ ~ 23 ₄ ... FIFO memory 31 ... Data extraction and output means 32 ... Validity determination means 33 (1) to 33 (4) ... Selection means

Claims (6)

[Claims] 1. A method for embedding invalid data in a received packet when a mismatch occurs at a boundary between new header information to be replaced according to a frame format and data of the packet, and valid for each byte of data. Means for setting an invalid flag; and invalid data removing means for arranging and outputting a plurality of bytes of valid data in a predetermined order while removing the invalid data based on the set flag. Network relay device. 2. When a mismatch occurs at a boundary between each new header information to be replaced according to a plurality of different frame formats and the same data of the packet, invalid data is embedded in the received packet. Means for setting a valid / invalid flag for each byte of data; and a frame format corresponding to aligning and outputting a plurality of bytes of valid data in a predetermined order while removing the invalid data based on the set flags. A network relay device comprising: invalid data removing means. 3. The network relay device according to claim 1, wherein the invalid data removing unit stores a plurality of byte data for each byte data including the invalid data; High-speed switching means for removing the invalid data while selecting an output path of the plurality of storage means based on the valid / invalid flag of the byte data, and aligning and outputting plural bytes of valid data in a predetermined order; A network relay device comprising: 4. The network relay device according to claim 1, wherein the invalid data removing unit stores a plurality of registers configured to store n bytes of byte-based data including the invalid data; Based on the valid / invalid flag of the byte unit data, the invalid data is removed while selecting the output of the register (n-1) from the register group having a plurality of stages, and the valid data of plural bytes is arranged in a predetermined order. And a selection means for arranging and outputting the same. 5. A header generating step for generating new header information according to a frame format for a received packet, and when a mismatch occurs between a boundary between the generated new header information and data of the packet. While embedding invalid data in the data embedding step of setting a valid / invalid flag for each byte unit data, removing the invalid data based on the set valid / invalid flag,
A byte alignment step of aligning and outputting a plurality of bytes of valid data. 6. A header generating step of generating new header information for each of a plurality of different frame formats with respect to a received packet, and determining a boundary between each of the generated new header information and the same data of the packet. While embedding invalid data in the event of inconsistency,
A data embedding step of setting a valid / invalid flag for each byte of data, a byte aligning step of removing the invalid data based on the set valid / invalid flag, and aligning and outputting a plurality of bytes of valid data respectively. A network relay method comprising: