JP4263115B2 - Data buffer control method and data relay apparatus - Google Patents

Description

  The present invention relates to a data buffer control method for communication data and a data relay device, and more particularly to a data buffer control method for communication data suitable for discard control of an Ethernet (registered trademark) frame based on a transfer delay time inside a bridge switch. And a data relay device.

  The layer 2 bridge switch, which is an Ethernet network connection device, checks the destination MAC address of the received Ethernet frame against the MAC address table held inside the device, and only the Ethernet frame is sent to the destination port corresponding to the matched MAC address. It has a mechanism to transfer. A large amount of data is being communicated on a network such as Ethernet, and there is a demand for speeding up Ethernet frame transfer processing of bridge switches.

  When the transfer process is performed by the bridge switch, the transfer speeds of the network of the reception source and the network of the transmission destination are not always the same. Further, when congestion occurs in the transmission destination network, the transfer to the transmission destination port in the bridge switch is not always performed at the same speed as that of the reception source port. In such a case, the transfer rate to the transmission destination port may be slower than the reception source port, and the transfer may be stagnant. If the bridge switch does not have an appropriate frame discard control function and the Ethernet frame transfer to the destination port is delayed, the Ethernet frame received after that will also be delayed. Will occur.

  The Ethernet frame received at the bridge switch input port is temporarily stored in the frame buffer. However, if the transfer to the destination output port is delayed or stagnant as described above, the received Ethernet frame data The amount may exceed the storage capacity of the frame buffer. In this case, subsequent Ethernet frame reception processing becomes impossible, so transmission of the Ethernet frame to the counterpart device that transmits the Ethernet frame by flow control using a back pressure or PAUSE frame is performed. The processing to stop is requested, and the network performance is degraded.

  Therefore, in the bridge switch, it is necessary to perform processing for discarding Ethernet frames that cannot be transferred normally even after a certain time. The upper limit value of the delay time from the reception of an Ethernet frame as a frame discarding reference to transmission is the IEEE std. Defined by the American Institute of Electrical and Electronics Engineers (IEEE). In 802.1D, it is standardized as a maximum bridge transfer delay (recommended value 1 second, maximum value 4 seconds). According to this standard, an Ethernet frame that has passed for a certain period of time in the bridge switch due to a delay or stagnation of the transfer is deleted because it is no longer required to be transferred.

  As conventional Ethernet frame discard control, there is a method using a time stamp. FIG. 8 shows an example of a conventional method using a time stamp. As shown in the figure, when the frame data input from the reception block 81 is sequentially written to the frame buffer 82, a time stamp is obtained for each frame and stored together with the frame data. Thereafter, when the frame is transferred to the transmission block 83, the passage delay time generated in the bridge switch is calculated from the difference between the time stamp acquired at the time of reception and the transmission time, and this passage delay time is the maximum bridge transfer delay time. The frame is discarded when the number exceeds.

  For example, Patent Document 1 discloses data discard control using a time stamp. The features of the method disclosed here are as follows: (1) When data (Ethernet frame or the like) input from the reception block is sequentially written to the reception buffer (frame buffer), time management data (time stamp or the like) is also written together. (2) Data in the receive buffer is transferred in a first-in first-out manner, and data whose transfer destination is not determined is returned to the end of the processing queue. (3) Time management data is checked when returning to the processing queue. Then, when a certain time has passed, the data is discarded.

  However, this method requires a time stamp storage area for each received Ethernet frame and a time stamp check circuit, which increases the circuit scale. Also, if the delay time is longer than the time stamp, the passage delay time cannot be determined. Therefore, the time stamp needs to be more accurate than the possible passage delay time, and the circuit scale increases further as the accuracy is increased. Cause. In addition, the discard processing based on the time stamp comparison of data stored in the frame buffer for a long time is performed in the order in which the data is transferred, so the data is immediately discarded when the maximum bridge transfer delay elapses. The frame buffer usage efficiency is poor.

Patent Document 2 discloses an improvement in packet discard control using a time stamp. FIG. 9 is a block diagram illustrating a configuration of a network connection device disclosed in Patent Document 2. The packet receiving unit 91 receives a packet from the connected segment,
The time stamp acquired from the time stamp processing unit 95 is stored in the buffer 94, and reception information such as a reception port or a destination address of the packet is passed to the routing processing unit 93. The routing processing unit 93 determines the transmission destination port of the packet from the reception information, and then passes transmission information such as the address of the buffer 94 in which the packet is stored to the delay management unit 96 of the transmission destination port.

  When the delay management unit 96 receives transmission information when there is a vacancy in the transmission queue 97 provided in the same port, the delay management unit 96 queues the transmission information as it is in the transmission queue 97. On the other hand, when the transmission information is received when there is no space in the transmission queue 97, after waiting for the transmission queue 97 to be free, the elapsed time from the time indicated by the time stamp of the packet to the current time is calculated. Calculated as delay time (residence time). If the delay time (dwell time) exceeds the maximum bridge transfer delay time of the network connection device, the packet is discarded. If not, the transmission information is sent to the transmission queue 97. To queue. Finally, based on the transmission information stored in the transmission queue 97, the packet transmission unit 92 reads a packet corresponding to the transmission information from the buffer 94, and transmits the packet to the segment connected to the transmission destination port. .

  By processing as described above, in the case of a packet whose transmission destination port is determined while there is a space in the transmission queue 97, the delay time calculation process is not performed, and the transmission information is immediately queued in the transmission queue 97. Transmission processing capability can be improved. However, if there is no vacancy in the transmission queue, it is necessary to wait for the vacancy in the transmission queue and then calculate the elapsed time from the time indicated by the time stamp of the packet to the current time as the passage delay time. For this purpose, the time stamp for each packet needs to be stored in the buffer together with the packet, which leads to an increase in circuit scale as in the case of Patent Document 1 described above. Alternatively, after waiting for the transmission queue to become empty, the packet passing delay time is calculated, so if there is no space in the transmission queue, the packet is not discarded even if the maximum bridge transfer delay time elapses. Usage efficiency is poor.

  Further, Patent Document 2 discloses an improvement in packet discard control that does not use a time stamp. The method disclosed here is as follows: (1) A packet buffer is divided, a reception area for storing packets received in a certain period, and a packet to be transmitted among packets received before the period is stored Specify the area and the discard area where the packets to be discarded are stored. (2) Cancel the designation of the area that has been the discard area at regular intervals, change the designation of the area that has been the reception area to the received area, and change all or part of the area that has been the received area to The designated area of the packet buffer is designated as the reception area other than the reception area, the received area, and the discard area before the area designation is changed. .

  Further, (3) when a packet is received, the packet is stored in the reception area designated as the reception area at the time of reception, a transmission destination port is determined based on the information of the packet, and the transmission destination port is set. In addition to queuing transmission information such as packet addresses in the provided transmission queue for sequentially reading information for packet transmission, (4) when the area designation is changed, the information is stored in the discard area Based on the discard information such as the packet address, the transmission information corresponding to the discard information is discarded from the transmission queue.

However, for a certain period of time, a buffer area that has been partitioned in advance is used as a receiving area. Therefore, when a large amount of frame data is temporarily received, all the data is stored in the receiving area allocated in advance. There is a possibility that it cannot be stored. On the other hand, even if the reception area is occupied by almost no data, the buffer usage efficiency deteriorates because the reception area is changed to the reception area after a certain period of time. End up.
JP-A-6-46064 Japanese Patent Laid-Open No. 10-13472

  As described above, in the conventional frame discard control using the time stamp, the time stamp storage area and the time stamp check circuit for determining that a predetermined time or more has elapsed are required due to the time stamp, which increases the circuit scale. There is a problem in terms of connection efficiency or use efficiency of the data buffer. Alternatively, in the conventional frame discard control that does not use a time stamp, the data buffer area is divided and the frame discard control is performed for each divided area, so that the data buffer cannot be used sufficiently efficiently.

  The present invention has been made against the background of the above circumstances, and is to enable efficient use of a data buffer by an efficient method and configuration.

A data buffer control method according to the present invention is a data buffer control method for temporarily storing communication data, the step (a) of storing communication data in the data buffer, and the data stored in association with the communication data. The step (b) of setting the aging data to the first state, and the aging data is in the first state when the first timing indicated by the timer indicating the timing at regular intervals is reached. In response to the step (c) of setting the aging data to the second state, and when the second timing indicated by the timer is reached after the first timing, as invalid the communication data not yet transmitted state in response to that in the second state, the communication data buffer In which a step (d) to open the band. Thus, efficient communication data discard can be performed without using a time stamp.

  The above-mentioned data buffer control method may include a step of storing other communication data acquired after the communication data in the communication data buffer area. As a result, the released buffer area can be used as a storage area for new communication data, and the use efficiency of the frame buffer can be improved.

  In step (d), the state data stored in association with the communication data and set to invalid state data indicating validity / invalidity of the communication data, and the communication data is invalid with reference to the state data. And a step of determining. Thus, the step of releasing the buffer area based on the validity / invalidity of the state data can be reliably performed.

In the step (b) , the aging data may be set to the first state in response to a storing process of the communication data in the data buffer. As a result, the time when the communication data is written to the data buffer can be set as the reference time for determining whether to discard the communication data, and the passage delay time can be appropriately determined.

In the step (b) , the aging data may be set to the first state in response to a reading process of the communication data from the data buffer . As a result, the time when the transfer to the transmission destination output port could not be performed can be used as the reference time for determining whether to discard the communication data, and the passage delay time can be appropriately determined.

  The aging data may represent two states of High and LOW. As a result, the amount of data necessary for storing aging data can be reduced.

  The communication data may be data transmitted from a plurality of output ports, and aging data associated with the communication data may be stored corresponding to each of the plurality of output ports. Thus, the present method can be effectively applied to a data relay device having a plurality of output ports.

  The communication data may be divided into a plurality of partial data and stored in the data buffer, and aging data and state data associated with each of the plurality of partial data may be stored. As a result, the next communication data can be written in divided data buffer units, and the use efficiency of the data buffer can be improved.

  The status data may be set invalid in response to the corresponding partial data being read from the data buffer. As a result, the data buffer area storing the communication data for which the transfer has been completed is immediately released, and new communication data can be taken in, so that the use efficiency of the data buffer can be improved.

On the other hand, the data relay device according to the present invention is a data relay device that relays acquired communication data, and temporarily associates the acquired communication data with the communication data stored in the data buffer. a storage unit for storing aging data Te, a timer for instructing the timing at regular intervals, Ru and a management unit. Further, the management unit (a) sets the aging data stored in the storage unit to a first state, (b) when the first timing indicated by the timer is reached, In response to aging data being in the first state, setting the aging data to a second state, and (c) at a second timing indicated by the timer after the first timing. When it arrives, in response to the aging data being in the second state, the communication data in the untransmitted state is invalidated and the buffer area for the communication data is released. With this configuration, communication data can be discarded without using a time stamp, so that the required circuit scale or processing time can be reduced.

  The storage unit stores state data indicating validity / invalidity of the communication data, and the state data is set to be invalid when the aging data is in the second state at the third timing. You may do it. Thus, the step of releasing the buffer area based on the validity / invalidity of the state data can be reliably performed.

  The storage unit stores aging data corresponding to each of a plurality of communication data, and sets the aging data stored in the storage unit to the second state at a predetermined interval instructed by the timer. You may make it do. As a result, communication data can be discarded without using a time stamp for a data buffer having an area for storing a plurality of communication data, so that the required circuit scale or processing time can be reduced.

  The storage unit stores state data indicating validity / invalidity of the communication data, and stores state data corresponding to the aging data set in the second state at the timing of the predetermined interval indicated by the timer. You may make it set invalid. Thus, the step of releasing the buffer area based on the validity / invalidity of the state data can be reliably performed.

  The management unit may set the aging data to the first state in response to a storage process of the communication data in the data buffer. As a result, the time when the communication data is written to the data buffer can be set as the reference time for determining whether to discard the communication data, and the passage delay time can be appropriately determined.

  The management unit may set the aging data to the first state in response to a data read process stored in a buffer area in which the communication data is stored. As a result, the time when the transfer to the transmission destination output port could not be performed can be used as the reference time for determining whether to discard the communication data, and the passage delay time can be appropriately determined.

  The storage unit stores status data corresponding to each of the plurality of output ports to which the communication data is transmitted, and in response to transmission of the communication data from one of the plurality of output ports, The state data corresponding to one output port may be set invalid. Thus, the present method can be effectively applied to a data relay device having a plurality of output ports.

  The data buffer divides and stores the communication data into a plurality of partial data, and the storage unit stores state data associated with each of the plurality of partial data, and stores one of the plurality of partial data. In response to the transmission, the state data associated with the one partial data may be invalidated to release the buffer area. As a result, the next communication data can be written in divided data buffer units, and the use efficiency of the data buffer can be improved.

  The present invention enables efficient use of a data buffer for communication data by an efficient method and configuration.

Embodiment 1 of the Invention
The Ethernet frame discard control according to the first embodiment of the present invention will be described with reference to FIGS. FIG. 1 schematically shows a partial configuration of a bridge switch according to the present embodiment. The reception block 1 has a function of transferring frame data received from the input port to the frame buffer 2. The transmission block 3 has a function of transmitting the frame data transferred from the frame buffer 2 to the output port.

  The frame buffer 2 is a shared memory that stores frame data received from the reception block 1. The frame buffer 2 is divided so that it can be stored in units of Ethernet frames. For example, the size of one area may be selected to be a size that can store 1518-byte data, which is the maximum size of an Ethernet frame. A larger size may be assigned in consideration of additional data such as a VLAN tag. However, if it is assigned more than necessary, the frame buffer utilization efficiency deteriorates, so an appropriate size may be selected according to the received data.

  The frame buffer management unit 4 manages the frame buffer 2 and controls frame data input to the frame buffer or frame data output from the frame buffer. The reception block 1 or the transmission block 2 inputs / outputs frame data to / from the frame buffer via the frame buffer management unit 4.

  The buffer data information register 5 stores control information associated with each frame data stored in the frame buffer 2. Specifically, the valid flag 6 and the aging flag 7 are stored for each frame data. Here, the valid flag 6 is an identifier indicating the validity of the frame data stored in the frame buffer 2, and is, for example, a 1-bit register that can be identified by a binary value of “High” indicating validity and “Low” indicating invalidity. Good. Here, “High” is a “1” value in the 1-bit register, and “Low” is a “0” value in the 1-bit register. The aging flag 7 is an identifier indicating that a certain time has elapsed since the frame data was stored in the frame buffer. For example, “High” indicates aging and “Low” indicates no aging. A 1-bit register that can identify presence or absence by binary values may be used.

  The buffer data information register 5 is managed by the register management unit 8. The register management unit 8 changes the data in the register 5 or passes the data in the register 5 to another configuration unit according to data or a request from another configuration unit. The aging timer 9 is a timer for giving an aging timing instruction for changing the aging flag 7 in the buffer data information register 5.

  The operation of the bridge switch according to the embodiment of the present invention will be described below with reference to the flowchart of FIG.

(1) Transfer from reception block 1 to frame buffer (write operation)
Frame data input from the input port to the reception block 1 is transferred to a free area of the frame buffer 2. If the write operation to the frame buffer 2 is normally performed (S201), a write end notification including necessary information such as address information is sent from the frame buffer management unit 4 to the register management unit 8. The register management unit 8 changes the buffer data information register 5 corresponding to the buffer area (S202). Specifically, the valid flag 6 is set to “High” (valid state), and the aging flag 7 is set to “Low” (no aging).

(2) Transfer from frame buffer 2 to transmission block 3 (read operation)
When the transmission block 3 can receive the frame data from the frame buffer 2, it makes a frame transfer request to the frame buffer management unit 4. The frame buffer management unit 4 determines, based on the information in the buffer data information register 5 acquired from the register management unit 8, whether there is a frame that can be transferred to the transmission block 3 that is making a data request in the area of the frame buffer 2. . (S203) If it exists, a read start notification is sent to the register management unit 8. The register management unit 8 sets the aging flag 7 corresponding to the buffer area in which the transfer frame is stored to “Low” (S204). The frame buffer management unit 4 performs data transfer to the transmission block 3. Here, by setting the aging flag 7 to “Low”, the aging is initialized at the frame transfer start time, and the stay time in the buffer can be strictly determined as the passage delay time.

  The frame buffer management unit 4 determines whether or not the frame transfer is completed (S205), and when it is completed, sends a transfer completion notification to the register management unit 8. In response to the notification, the register management unit 8 changes the valid flag 6 corresponding to the buffer area in which the transferred frame is stored to “Low” (S206). As a result, new frame data can be immediately written to the buffer area.

(3) Aging (Data Invalidation) Procedure The register management unit 8 determines the specified time by the aging timer 9 (S207). By acquiring the aging timing instruction from the aging timer 9, the register management unit 8 determines that the designated time is reached. The register management unit 8 determines the valid flag 6 and the aging flag 7 of the buffer data information register 5 at the designated timing (S208). If the aging flag 7 is “Low” at the designated timing, the register management unit 8 changes the value of the aging flag 7 to “High” (S209). When the aging flag 7 is “High” and the corresponding valid flag 6 is “High” at the designated timing, the register management unit 8 sets the valid flag 6 to “Low” (S210). Note that the change of the buffer data information at the designated timing is executed for the buffer data information corresponding to all the frame data. As a result, the frame data stored in the frame buffer 2 is invalidated, the buffer area is released, and the frame data can be used for writing new frame data.

  Here, the change timing of the buffer data information in the buffer data information register 5 based on the aging timing given by the aging timer 9 will be described with reference to the timing charts of FIGS. 3 (a) and 3 (b).

  FIG. 3A shows a case where the frame data transfer to the transmission block 3 is completed before the frame data is invalidated by aging. In the figure, an arrow indicated as “Write” indicates a flag change due to data being written to the frame buffer (transfer of frame data to the frame buffer 2), and an arrow indicated as “Read” indicates that data is transferred from the frame buffer. The change of the flag due to reading (frame data transfer from the frame buffer 2) is shown. As the frame data is transferred to the frame buffer 2, the aging flag is changed to “Low” and the valid flag is changed to “High”. Thereafter, the aging flag is changed to “High” when the aging timing is first reached. Subsequently, the aging flag is changed to “Low” again at the start of reading frame data, and then the valid flag is changed to “Low” indicating data invalidity at the end of reading. Accordingly, it can be determined that the frame buffer area is released and the next frame data can be written.

  FIG. 3B shows a process in which data is invalidated by aging. The aging flag is changed to “Low” and the valid flag is changed to “High” by the write process. Thereafter, the aging flag is changed to “High” when the aging timing is first reached. In the figure, since the next aging timing is reached without data being read out thereafter, the valid flag is changed to “Low” at this time. As a result, the frame data stored in the frame buffer area is invalidated, and it can be determined that new data can be written in this area.

  Assuming that the aging timing interval in the Ethernet frame discard control according to this embodiment is Te, the time from when the frame data is written to the frame buffer area until the invalidation of the frame data by the aging procedure is from Te to After 2 Te. For example, when Te is 4 seconds, the frame data is discarded after 4 to 8 seconds.

  It is possible to reduce the fluctuation of the time until the frame is discarded by reducing the aging timing and increasing the aging age. For example, the aging timing interval Te is set to 1 second, the aging flag is changed from the above-described two stages (High and Low) to four stages, the aging flag is aged sequentially every time the aging timing elapses, and the aging flag is the fourth stage. If the invalidation is performed by setting the valid flag to “Low” when the aging timing is reached, the frame data is discarded after 4 Te to 5 Te, that is, after 4 to 5 seconds. From the viewpoint of circuit scale, it is preferable that the number of bits of each flag is small, particularly 1-bit data.

  As described above, in the control method according to the embodiment of the present invention, by appropriately setting the interval of the aging timing specified by the aging timer according to the allowable maximum bridge transfer delay, compared with the case of using the time stamp. Frame discard control with a small circuit scale.

  For example, when a conventional time stamp is used and there are 20 frames of time stamps that can be identified up to 5 minutes at 1 millisecond intervals, a total storage area of 380 bits is required for the time stamp. On the other hand, with the configuration according to the present embodiment using one aging timer, 1-bit valid flag, and 1-bit aging flag, only a 40-bit storage area is required, so that the storage area can be reduced and the time is further increased. A circuit for delay time calculation using a stamp is also unnecessary.

  In the control method according to the embodiment of the present invention, the frame data stored in the buffer area after the set passage delay time elapses immediately without waiting for the transfer processing timing to the transmission block. Since the buffer area is released, it is possible to improve the use efficiency of the frame buffer and reduce the passage delay time in the bridge switch.

  Furthermore, in the control method according to the embodiment of the present invention, whether or not new frame data can be received in the frame buffer area is determined based on the validity flag, so that the buffer area is fixedly secured as a reception area in advance. Therefore, it is excellent in durability when a large amount of frame data is temporarily received. In addition, since new frame data can be received unless new frame data is received and the valid flag is set to “High”, compared to the conventional method in which the received area is uniformly received at regular intervals, The use efficiency of the frame buffer can be improved.

Embodiment 2 of the Invention
The Ethernet frame discard control according to the second embodiment of the present invention will be described with reference to FIGS.

  FIG. 4 shows a configuration diagram of the bridge switch according to the present embodiment. The embodiment of the present invention is different from Embodiment 1 in that the frame buffer 2 is divided into fixed length units. The data frame is stored in one fixed length unit or divided into a plurality of fixed length units. In addition to the valid flag 6 and the aging flag 7, link information 10 is held in the buffer data information register 5. The link information 10 indicates the address of the fixed length area where the next divided portion is stored when one frame is divided into a plurality of fixed length areas and stored.

  In FIG. 4, as an example, the frame buffer 2 is divided into fixed-length areas of 512 bytes. In this case, the frame data received from the receiving block 1 is stored using one to three planes according to the frame length (64 bytes to 1518 bytes) (the plane is a unit for counting each fixed length area). is there). The buffer data information register 5 has information corresponding to each fixed-length divided area. Since other configurations are the same as those of the first embodiment, description thereof will be omitted.

  The frame discard control according to the embodiment of the present invention differs from the first embodiment of the present invention in that the frame transfer from the frame buffer 2 to the transmission block 3 is (1) the transfer of one fixed length buffer is completed. At this point, the valid flag corresponding to the target fixed-length buffer area is changed to “Low”. (2) If there is a data link destination (if two or three sides are used), then Similarly, data is transferred from the fixed length area of the link destination to the transmission block 3, and the valid flag is changed to "Low" at the end of the transfer.

  The change timing of the buffer data information in the buffer data information register 5 based on the aging timing given by the aging timer 9 in this embodiment will be described with reference to the timing charts of FIGS. 5 (a) and 5 (b).

  FIG. 5 (a) shows a case where frame data transfer to the transmission block is completed before the data is invalidated by aging, and corresponds to FIG. 3 (a) in the first embodiment of the invention. It is.

  The arrows indicated as Write 1/2 in the figure indicate data change caused by writing data for one frame divided into two fixed length areas. In response to data writing (data transfer to the frame buffer), the aging flags corresponding to the two fixed length areas (first side, second side) are changed to “Low”, and the valid flag is changed to “High”. Thereafter, when the aging timing is first reached, the aging flags of the two fixed-length areas are changed to “High”. Subsequently, at the start of reading frame data, the aging flag of the fixed length area (corresponding to the first surface in the figure) read first is changed to “Low”, and the valid flag is changed to “Low” at the end of reading. (Arrow indicated as Read 1 in the figure). At this point, new data can be taken into this fixed-length area.

  Furthermore, the aging flag corresponding to the fixed-length area on the second side is changed to “Low” by the start of reading data of the next fixed-length area (corresponding to the second side in the figure), and the valid flag is set at the end of reading. It is changed to “Low” (an arrow indicated as Read 2 in the figure). At this point, new data can be taken into this fixed-length area.

  FIG. 5B shows a process in which data is invalidated by aging, and corresponds to FIG. 3B in the first embodiment of the invention.

  The arrows indicated as Write 1/2 in the figure indicate data change caused by writing data for one frame divided into two fixed length areas. In response to data writing (data transfer to the frame buffer), the aging flags corresponding to the two fixed length areas (first side, second side) are changed to “Low”, and the valid flag is changed to “High”. Thereafter, when the aging timing is first reached, the aging flags of the two fixed-length areas are changed to “High”. In the figure, since the next aging timing is reached without data being read out thereafter, the valid flag corresponding to the two fixed-length areas (first and second faces) is changed to “Low” at this time. . As a result, the data stored in the buffer area is determined to be invalid, and new data can be written to the corresponding fixed-length area. Here, if the aging timing comes after the frame data of the first page is read out, it can be invalidated by changing the valid flag of the frame data of the second page to “Low”, and a new data frame can be written. It can be.

  As described above, in the control method according to the embodiment of the present invention, new data can be written one after another from the fixed-length buffer area where the transfer has been completed without waiting for the transfer of the entire one frame data. Can be further improved and the bridge switch transit delay time can be reduced.

  For example, when waiting for the transfer of a 1518-byte Ethernet frame and requiring 10 nanoseconds to transfer an 8-byte data, dividing the frame buffer into 512-byte units will start writing to the next frame buffer. Can be accelerated by up to about 1.2 microseconds compared to waiting for the transfer of the entire frame.

Embodiment 3 of the Invention
When a unicast frame (Ethernet frame transferred to only one port) is required in a switch having multiple input / output ports, "transfer destination information" is added to the buffer data information register to transfer the frame data. In some cases, control may be performed so as to perform transfer processing for the output port indicated in the transfer destination information.

  FIG. 6 shows the configuration of the buffer data information register in this embodiment. By configuring in this way and controlling to set the valid flag to “Low” when the transfer process for the output port indicated in the transfer destination information is completed, new writing to the frame buffer area is possible. Can be judged.

  Even if the transfer to the output port indicated by the transfer destination information is not completed, if the aging operation is performed by the aging timer, the valid flag bit is set to “Low” to write to the frame buffer area. Is possible.

Embodiment 4 of the Invention
If you need to transfer a multicast frame (the same Ethernet frame needs to be transferred to multiple ports) in a bridge switch with multiple input / output ports, make sure that you have a valid flag according to the number of output ports or transmission blocks. The bit of the valid flag corresponding to the output port to be transmitted at the time of writing to the frame buffer may be set to “High”.

  The configuration of the buffer data information register in this embodiment is shown in FIG. The data representing each transfer destination and the corresponding valid flag are represented by the same bit string. An aging flag is stored in association with each transfer destination and valid flag. By configuring in this way and controlling to change the bit of the corresponding valid flag to “Low” when the transfer to each transmission block is completed, it is effective when all the necessary transfer to the transmission block is completed. All the bits of the flag become “Low”, and it can be determined that new writing to the frame buffer area is possible.

  Even when transfer to all necessary transmission blocks is incomplete, the aging flag is changed by the aging operation by the aging timer, and all bits of the valid flag are set to “Low” according to the aging flag. Writing into the frame buffer area becomes possible.

  In this specification, an example of a bridge switch that transfers an Ethernet frame is described as a preferred mode. However, the present invention can be applied to other data transfer systems. In addition, those skilled in the art can easily change, add, and convert each element of the above-described embodiment within the scope of the present invention.

It is a block diagram of the bridge switch used for Ethernet frame discard control concerning Embodiment 1 of invention. It is an operation | movement flow of the Ethernet frame discard control concerning Embodiment 1 of invention. It is a schematic diagram of the aging operation | movement concerning Embodiment 1 of invention. It is a block diagram of the bridge | bridging switch used for Ethernet frame discard control concerning Embodiment 2 of invention. It is a schematic diagram of the aging operation | movement concerning Embodiment 2 of invention. It is a block diagram of the buffer data information register concerning Embodiment 3 of invention. It is a block diagram of the buffer data information register concerning Embodiment 4 of invention. It is a block diagram of the data communication apparatus used for the conventional Ethernet frame discard control. It is a block diagram of the data communication apparatus used for the conventional Ethernet frame discard control.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Reception block 2 Frame buffer 3 Transmission block 4 Frame buffer management part 5 Buffer data information register 6 Valid flag 7 Aging flag 8 Register management part 9 Aging timer

Claims (17)

A data buffer control method for temporarily storing communication data,
(A) storing communication data in a data buffer;
(B) setting aging data stored in association with the communication data to a first state;
(C) When the first timing indicated by the timer that indicates the timing at regular intervals has been reached, the aging data is set to the second state in response to the aging data being in the first state. Step to set to,
(D) When the second timing indicated by the timer is reached after the first timing, the communication data in an untransmitted state in response to the aging data being in the second state Disabling and releasing the buffer area of the communication data;
A data buffer control method.   The data buffer control method according to claim 1, further comprising a step of storing other communication data acquired after the communication data in a buffer area of the communication data. The step (d) is stored in association with the communication data, and the state data indicating validity / invalidity of the communication data is set to invalid,
Referring to the status data and determining that the communication data is invalid;
The data buffer control method according to claim 1 or 2 , comprising: The said step (b) WHEREIN: The said aging data is set to a said 1st state in response to the memory | storage process to the said data buffer of the said communication data, The any one of Claims 1-3 . Data buffer control method. In the step (b), the aging data, the response to the read processing of the communication data from the data buffer is set in the first state, according to any one of claims 1 to 4 Data buffer control method.   The data buffer control method according to claim 1, wherein the aging data represents two states of High and Low. The communication data is data transmitted from a plurality of output ports,
The data buffer control method according to claim 1, wherein aging data associated with the communication data is stored corresponding to each of the plurality of output ports. The communication data is divided into a plurality of partial data and stored in the data buffer,
Aging data and state data associated with each of the plurality of partial data are stored.
The data buffer control method according to claim 3.   9. The data buffer control method according to claim 8, wherein the status data is set invalid in response to the corresponding partial data being read from the data buffer. A data relay device that relays acquired communication data,
A data buffer for temporarily storing the acquired communication data;
A storage unit for storing aging data in association with communication data stored in the data buffer;
A timer that directs timing at regular intervals;
A management department,
The management unit
Setting the aging data stored in the storage unit in the first state,
When reaching the first timing instructed by the timer, that the aging data sets the aging data in response to in the first state to the second state, and the first when it reaches the second timing indicated by the timer after the timing, together with the aging data is to disable the communication data not yet transmitted state in response to that in the second state, the A data relay device that executes the release of a buffer area for communication data. The storage unit stores state data indicating validity / invalidity of the communication data,
In the second timing, wherein when the aging data is in the second state, said state data is set to invalid, the data relay apparatus according to claim 10. The storage unit stores aging data corresponding to each of a plurality of communication data,
The data relay device according to claim 10 or 11 , wherein the management unit sets the aging data stored in the storage unit to the second state at a predetermined interval instructed from the timer. The storage unit stores state data indicating validity / invalidity of the communication data,
The management unit, at the timing of the predetermined intervals indicated from the timer is set to disable status data corresponding to the age data set in said second state,
The data relay device according to claim 10 . The data relay device according to any one of claims 10 to 13, wherein the management unit sets the aging data to the first state in response to a storage process of the communication data in the data buffer. . The data relay device according to any one of claims 10 to 14, wherein the management unit sets the aging data to the first state in response to a process of reading the communication data from the data buffer. . The storage unit stores state data corresponding to each of a plurality of output ports to which the communication data is transmitted,
The data relay device according to claim 11, wherein in response to transmission of the communication data from one of the plurality of output ports, status data corresponding to the one output port is set to be invalid. The data buffer stores the communication data divided into a plurality of partial data,
The storage unit stores state data associated with each of the plurality of partial data;
In response to one transmission of the plurality of partial data, the state data associated with the one partial data is invalidated and the buffer area is released.
The data relay device according to claim 11.

Images