JP2012049835A - Transmission device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To avoid unnecessary bands generated by adding a frame length due to padding processing for fragment frames smaller than a minimum frame length in fragmentation of signals.SOLUTION: A packet signal from which a header is removed is divided by a fragment length to generate fragments. When a frame length of a part including the end opposite from the header in the packet signal is shorter than the sum of the fragment length and a threshold, this part is processed so as not to be divided, and the packet signal is divided into multiple fragments.

Description

  The present invention relates to a fragment scheme process in which an input frame is divided into several frames in a packet communication signal transmission apparatus typified by Ethernet (registered trademark) defined in IEEE802.3.

    In recent years, broadband lines such as the Internet have become widespread in homes, and the demand for lines has been increasing, mainly for IP traffic. Inexpensive Ethernet is rapidly gaining popularity in the market.

    The Ethernet signal is data representing a no-signal state called IFG (Inter Frame Gap) between MAC (Media Access Control) frames defined in IEEE802.3 (Non-patent Document 1) and MAC frames (at least 12 bytes or more) ). MAC is a protocol belonging to the second layer (layer 2) of the OSI reference model, and its role is to store the protocol and data of the third layer (layer 3) and higher of the OSI reference model in the data area of the MAC frame. The stored layer 3 or higher protocol is reliably transmitted to the target terminal. Like the normal Ethernet frame shown in 1) -1 of Fig. 1, the MAC frame has a preamble (8 bytes) indicating the head of the MAC frame and a destination address (6 bytes) indicating the MAC address of the destination terminal of the MAC frame. A source address (6 bytes) indicating the MAC address of the terminal transmitting the MAC frame, a Type / Length (2 bytes) indicating the length of the MAC frame, a data area (variable length 46 to 1500 bytes), and It consists of a checksum value (4 bytes). Normally, the portion excluding the preamble is counted in the effective frame range, and the frame length is 64 bytes to 1518 bytes. 2) of FIG. 1 shows an Ethernet frame in which the frame length is extended to 1518 bytes or more while keeping the data format of the Ethernet frame. Such an Ethernet frame is called a jumbo frame. Since jumbo frames can reduce the ratio of IFG (no signal state) than normal frames, the transmission path band can be used more efficiently than when normal frames are used.

Furthermore, the inefficiency in equipment and maintenance is desired by coexisting a backbone network based on the existing SDH / SONET technology and a backbone network based on the new IP / Ethernet technology. FIG. 2 shows a frame encapsulating original data such as an SDH / SONET signal and an IP / Ethernet frame. To eliminate the above inefficiency, original data such as SDH / SONET signals and IP / Ethernet frames are encapsulated with the same standard headers and footers, and then consolidated into a frame communication based backbone network. It is intended. In particular,
Represented by T-MPLS (Transport-MPLS) technology. An apparatus that generates such an encapsulated frame is hereinafter referred to as an encapsulating apparatus.

IEEE802.3 IETF RFC4623 ITU-T Y.1415

  This encapsulation device is required to have functions such as jumbo frame support and bandwidth management. Hereinafter, a jumbo frame is a frame of a predetermined byte length or more, and particularly refers to a frame having a frame length of 1518 bytes or more from the relationship of 1518 bytes, which is the maximum length of an ether frame specified by IEEE802.3. To do. At present, there are cases where a network composed of encapsulation devices that support jumbo frames and a network composed of encapsulation devices that do not support jumbo frames coexist. FIG. 3 shows an example of a network composed of encapsulation devices that support and do not support jumbo packets. Fig. 3 shows a network A that supports jumbo frames between the client device 100-1 and the client device 100-2 (a network having a maximum transmission unit corresponding to a frame of a predetermined byte length or more. Maximum transmission of a communication line excluding a header) Unit (Maximum Transmission Unit, MTU) means 1518 bytes or more), Network B that does not support jumbo frames (Network with the maximum transmission unit corresponding to frames less than the specified byte length. The MTU of the communication line excluding the header is less than 1518 bytes. Meaning that there is a network C that supports jumbo frames, and each network is connected by the same transmission standard. When sending a jumbo frame from the client device 100-1 to the client device 100-2, if nothing is done, the network B does not support the jumbo frame, so the edge node 200-3 discards the frame. There are challenges.

  As a technique used in such a case, for example, there is a fragmentation technique defined in IETF RFC4623 (Non-Patent Document 2) and ITU-T Y.1415 (Non-Patent Document 3). In the fragmentation technique, a long frame is divided by a predetermined length. If the last fragment of the division is less than the specified minimum frame length, padding processing is performed to add “0” data until the minimum frame length is reached. An example is shown in FIG. 4 shows a case where a jumbo frame having a frame length of 1522 bytes is divided, and 2) of FIG. 4 shows a case where a portion excluding the header of the jumbo frame is divided in units of 1512 bytes. Here, a header is added to each divided fragment. In 2) of FIG. 4, the last fragment frame is divided into 10 bytes. Here, since 10 bytes is less than the minimum frame length of 64 bytes of Ethernet, padding processing for 54 bytes is performed. After the frame is divided in this way, for example, in ITU-T Y.1415, the IWF header shown in 1) of FIG. 5 is added. FIG. 5 2) shows the use of each field of IWF. Here, it is defined that the presence / absence of the execution of the fragment and its state are written in the FRG area, and that the frame length of the padded frame is written in the Length Field.

  A case where the jumbo frame of 1) in FIG. 4 is input to the network having the configuration of FIG. 3 in the case of using such a fragmentation technique will be described. The jumbo frame output from the client device 100-1 passes through the network A as it is and is fragmented at the edge node 200-3, resulting in a state as shown in 2) of FIG. The frame in this state is transmitted on the network B, and fragment coupling is performed at the edge node 200-4 to return to the state of 1) in FIG. Thereafter, the jumbo frame is received by the client apparatus 100-2 via the network C. Here, from the viewpoint of bandwidth, since padding is not a signal that is originally desired to be transmitted, a wasteful bandwidth is generated. In other words, in the above fragmentation technique, since a fragment frame that is less than the minimum frame length is padded with “0” data until the minimum frame length is reached, it is wasted by adding the frame length by adding “0” data. There is a problem that a large band is generated.

  In the transmission apparatus, when signal fragmentation is performed, the last fragment is determined to be divided / non-divided using a specified minimum frame length as a threshold value, and when it is determined to be non-divided, padding processing to the last fragment is performed. To avoid.

  The transmission apparatus according to the present invention, as an example, generates a fragment by measuring the frame length of a portion excluding the header of the packet signal and dividing the portion excluding the header of the packet signal by the first fragment length. When the frame length of the portion including the other end that is not the header side of the packet signal is less than the sum of the first fragment length and the threshold, the packet signal is processed so as not to be divided, and the packet signal is divided into a plurality of fragments. A first fragment processing unit, and a header adding unit that adds a header to the plurality of fragments.

  Here, when a portion excluding the header of the packet signal is divided by a second fragment length to generate a fragment, and a portion including the other end of the packet signal that is not the header side is less than the second fragment length A second fragment processing unit that adds data to generate a fragment of a minimum frame length according to the standard that the packet signal conforms to, and converts the packet signal into a plurality of fragments; the first fragment processing unit; and the second fragment processing And a selector that transmits the packet signal to one of the first fragment processing unit and the second fragment processing unit based on the selection of the method determination unit. Further, the header adding unit may be provided by either the first fragment processing unit or the second fragment processing unit. It may be imparted to the header into a plurality of fragments to be.

  When performing fragmentation of a packet communication signal, it is possible to eliminate the padding process and avoid a bandwidth shortage. In addition, since the padding process is not performed, it is possible to omit the process of writing the frame length in the IWF header of the fragment frame subjected to the padding process. Therefore, fragment processing can be realized with a simpler configuration.

Diagram showing Ethernet frame format The figure which shows the frame in an encapsulation device The figure which shows an example of the network comprised with the encapsulating device of jumbo frame correspondence and non correspondence Diagram showing an example of the fragment method Diagram showing IWF header The figure which shows the fragment method by Example 1 The figure which shows the flowchart of Example 1. The figure which shows an example of the circuit in Example 1. The figure which shows an example of the data band of Example 1. The figure which shows the flowchart of Example 2. The figure which shows an example of the circuit in Example 2. The figure which shows an example of the data band of Example 2.

  Embodiments of the present invention will be described below with reference to the drawings. Here, the signal input to the transmission apparatus may be an Ethernet signal or a packet signal to which an MPLS header is added.

  FIG. 6 shows a conceptual diagram of fragment processing according to the first embodiment. Here, the frame length of the part excluding the header of the input packet communication signal (frame) is L [Byte], the fragment length setting value to be divided into equal intervals is A [Byte], and the final fragment setting threshold is B [ Byte]. Here, B is the minimum frame length according to the standard followed by the input packet communication signal (frame). In this embodiment, the L frame is divided as A-length fragments (portion 601 in FIG. 6). And when the remaining frame length including the other end that is not the header side is A to 2A, it is determined whether the remaining frame length is less than or equal to A + B, and if it is less than A + B, The remaining frame length is output as it is, and if A + B or more, A and the remaining frame length fragment are output. That is, when the portion 602 in FIG. 6 is smaller than the set threshold B [Byte], the portion 602 in FIG. 6 and the adjacent portion are not divided.

  FIG. 7 shows a flowchart of the system of the first embodiment. FIG. 8 shows the configuration of the transmission apparatus (edge node). When a packet communication signal (frame sequence) is input to the transmission apparatus (edge node), the frame length measurement unit 500 measures the frame length of the portion excluding the header (A02). Thereafter, the fragment processing unit 501 determines whether the frame length of the portion excluding the header is greater than or less than the set value A + B (A03). If the frame length L [Byte] of the part excluding the header is greater than or equal to the set value A + B, the frame is divided into fragment frame lengths A (A04), and the frame that is not the fragment frame The frame length LA [Byte] is used again to determine whether the set value is A + B or less. This is repeated until the frame length of the non-fragment frame becomes less than A + B [Byte]. When the frame length of the non-fragment frame is less than A + B [Byte], the fragmentation process is completed without dividing the frame including the other end that is not the header side. After the fragmentation process is completed, a header is added to each fragment frame by the header adding unit 502 (A05).

  In this embodiment, the data bandwidth when the MTU is set to A + B-1 [Byte] is qualitatively as shown in FIG. The case according to the present embodiment is shown as the present method, and the band of the original frame (broken line) and the band of the conventional fragment method (dotted line) are shown as comparison targets. For comparison, the bandwidth of the original frame refers to the bandwidth of the frame when no fragment is virtually executed. The conventional fragment scheme for padding processing is when the MTU is the fragment length. The reason why the conventional fragment scheme band and the present scheme band are lower than the original band is that the band of the header portion of the fragment frame increases.

  A portion where the conventional fragment scheme band and the present scheme band suddenly become low indicates that fragmentation has been performed. This method eliminates bandwidth reduction due to padding performed immediately after fragment division performed in the conventional fragment method.

  In the fragment scheme of the first embodiment, when considering the MTU of the transmission system, it is necessary to make the fragment division length A smaller than the conventional padding scheme. In this case, when the MTU of the transmission system is D [Byte], the maximum fragment length frame length of the first embodiment is specified by fragment division length + minimum frame length-1 (A + B-1 [Byte]). Therefore, A = D−B + 1, and the maximum fragment frame length of the conventional method is defined only by the fragment division length (C [Byte]), and thus D = C. Therefore, since the fragment division length differs between the first embodiment and the conventional method, the data length of the frame length between the division point of the first embodiment and the division point of the conventional method is wider in the conventional method than in the first embodiment. . This band is small compared to the padding band, but in order to deal with system convenience more flexibly, in Example 2, by combining Example 1 and the conventional method according to the input frame length, All the frame lengths are set so that the data bandwidth is higher than that of the conventional method.

Therefore, in this embodiment, in addition to the transmission apparatus of the first embodiment, a fragment scheme determination unit is added so that the conventional padding processing scheme and the first embodiment can be used properly. Thus, by selecting and responding to the transmission apparatus input frame length that requires padding and the transmission apparatus input frame length that does not require padding in the first embodiment, the data bandwidth is reduced by the amount of padding for all frame lengths. It is possible to have more widely. Specifically, the frame length L [Byte] of the portion excluding the header of the input packet communication signal (frame), the division number n, and the fragment length setting value to be divided along the method of the first embodiment are A [Byte ], When the final fragment setting threshold (= standard minimum frame length) is B [Byte], and the fragmentation fragment length setting value of the conventional padding processing method is C [Byte]
(Equation 1)
nx C <L <nx C + B
When satisfying the above, fragmentation is performed by the method of the first embodiment.

  In the case of the present embodiment, it is possible to have a band equal to or higher than the conventional padding method while reducing the tightness of the band.

  FIG. 10 shows a flowchart of the fragment method according to the second embodiment. FIG. 11 shows the configuration of the transmission apparatus (edge node). Here, the input frame length L [Byte], the number of divisions n, the fragment length setting value to be divided along the method of the first embodiment is A [Byte], and the final fragment setting threshold (= the minimum frame length in the standard) is B [Byte], and the fragment fragment length setting value for the conventional padding method is C [Byte].

  When a packet communication signal (frame sequence) is input to the transmission device (edge node), the frame length measurement unit 1000 measures the frame length of the portion excluding the header (B02). Thereafter, the fragment method determination unit 1010 determines whether the relationship of nx C <L <nx C + B is satisfied for the frame length, and selects the fragment method (B03). If the relationship of nx C <L <nx C + B is satisfied, the fragment method determination unit 1000 controls the selector in the subsequent stage, sends the frame to the first fragment processing unit 1001, and the fragment method of the first embodiment Perform the fragment along. If the relationship of nx C <L <nx C + B is not satisfied, the fragment method determination unit 1000 controls the selector in the subsequent stage and sends the frame to the second fragment processing unit 1002, and the conventional padding shown below Perform fragment of processing mode. That is, first, it is determined whether L is greater than or equal to C (B04). In this case, fragmentation is performed with the frame length C (B05), and it is determined again whether the remaining frame length is greater than or equal to C. This is repeated until the frame length becomes less than C. When the frame length is less than C, “0” data is added until the minimum frame length is reached, and a fragment of the minimum frame length according to the standard that the packet communication signal follows is generated. In any case, the respective fragment frames are sent to the header assigning unit 1003 via the selectors subsequent to the first fragment processing unit 1001 and the second fragment processing unit 1002, and the headers are assigned to the fragment frames (B08). And output from the transmission device.

  In this embodiment, the data bandwidth when the MTU is set to A + B-1 [Byte] and C [Byte] is qualitatively as shown in FIG. When padding is required, the method of the first embodiment is adopted, so that it is possible to set the bandwidth for all frame lengths to be higher than that of the conventional padding method.

Claims (8)

A measurement unit that measures the frame length of the part excluding the header of the packet signal;
A fragment is generated by dividing a portion excluding the header of the packet signal by a first fragment length, and a frame length of a portion including the other end of the packet signal that is not the header side is the first fragment length and a threshold value A first fragment processing unit for processing so as not to divide the portion when the sum is less than the sum, and converting the packet signal into a plurality of fragments;
A transmission apparatus comprising: a header adding unit that adds a header to the plurality of fragments.   The first fragment processing unit, for each division by the first fragment length, whether the frame length of the portion including the other end of the packet signal that is not the header side is less than the sum of the first fragment length and the threshold The transmission apparatus according to claim 1, wherein it is determined whether or not.   The transmission apparatus according to claim 1, wherein the threshold is a standard minimum frame length that the packet signal complies with.   The transmission apparatus according to claim 1, wherein the threshold value is a minimum frame length of Ethernet.   The transmission apparatus according to claim 1, wherein the packet signal is an Ethernet signal.   The transmission apparatus according to claim 1, wherein the packet signal is a packet signal to which an MPLS header is added. Generate a fragment by dividing the portion of the packet signal excluding the header by the second fragment length, and add data when the portion of the packet signal including the other end that is not the header side is less than the second fragment length Generating a fragment of the minimum frame length according to the standard that the packet signal complies with, and converting the packet signal into a plurality of fragments;
A method determination unit that selects one of the first fragment processing unit and the second fragment processing unit;
A selector that transmits the packet signal to either the first fragment processing unit or the second fragment processing unit based on the selection of the method determination unit;
The transmission apparatus according to claim 1, wherein the header adding unit adds a header to a plurality of fragments sent from either the first fragment processing unit or the second fragment processing unit.   The method determining unit is nx C <L <nx C + B, where L is a portion excluding the header of the packet signal, n is the number of divisions, B is the threshold, and C is the second fragment length. 8. The first fragment processing unit is selected if the condition is satisfied, and the second fragment processing unit is selected if nx C <L <nx C + B is not satisfied. Transmission equipment.

Images