JP2012124700A - Packet switch - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a packet switch capable of securing high reliability in providing a function by a server.SOLUTION: A packet switch includes: a switch part including multiple ports to input/output packets; and a routing table. The switch part receives a first message including port information and an address, which indicate at least one of the multiple ports, via one of the multiple ports. The packet switch includes: a buffer part connected to the switch part via the first port among the multiple ports and receiving and storing a packet which is output from the first port; and a control part for setting the routing table thereby to output a packet from the first port, which is received by a port other than the port indicated by the port information included in the first message among the multiple ports and defines an address included in the first message as a destination address, when the first message is received.

Description

  The present invention relates to a packet switch for transferring a packet.

  At present, various servers are providing various functions to a large number of users via a network. As the number of users has increased and the importance of functions provided has increased, interest in stability and continuity of function provision has increased, and high reliability of servers is required. It is becoming.

  Therefore, when multiple servers are connected by packet switches and the software and hardware of the operation server that provides the function are replaced during normal operation, or when a failure occurs in the operation server, the function is provided. Switch the server from the active server to the backup server. Thereby, the time when provision of a function is interrupted can be shortened.

  In addition, when a failure occurs on the primary server, the server group is managed to continue to provide functions such as which software to be delegated to which server among the multiple backup servers. Technology for this is also considered. Such a technique is disclosed in Patent Document 1, for example.

  In the technology disclosed in Patent Document 1, in a system that provides functions by a plurality of servers, which process is executed by which server depending on the cause of the failure when a failure is detected Is managing. Thereby, the time for which the provision of the function is interrupted can be shortened as much as possible.

  Here, an example of operation | movement when switching the server which provides a function is demonstrated. Here, a case where the software and hardware of the operation server are systematically replaced will be described.

  FIG. 5 is a diagram for explaining an example of an operation when switching a server that provides a function.

  A packet switch 100 shown in FIG. 5 includes a switch unit 101 having ports 11a to 11c for inputting and outputting packets, a control unit 102, and a routing table unit 103. The servers 20 and 30 are connected to the network 50 via the packet switch 100.

  The routing table unit 103 stores a routing table. The routing table stores a port that receives a packet and a port that outputs the packet among the ports 11a to 11c in association with the destination address of the packet.

  When a packet is received at each of the ports 11a to 11c, the switch unit 101 determines a port to output the packet by referring to the routing table from the received port and the destination address of the packet. Then, the switch unit 101 outputs the packet from the determined port.

  The control unit 102 controls the operation of the entire packet switch 100. Specifically, for example, setting and updating of the routing table and state management of the switch unit 11 are performed.

The operation for switching servers in the configuration shown in FIG. 5 is as described below. Here, the server 20 is an operation server and the server 30 is a backup server.
(Procedure a) The server 20 activates the server 30 while providing functions to terminals connected to the network 50.
(Procedure b) The server 20 copies various environmental data to the server 30 while continuing to provide the functions.
(Procedure c) The server 20 stops providing functions to terminals connected to the network 50.
(Procedure d) The server 20 copies the status data regarding the service being processed to the server 30.
(Procedure e) The server 20 shifts the server 30 to the operating state.
(Procedure f) The control unit 102 of the packet switch 100 sets a routing table so that a packet transmitted from the network 50 to the server 20 is transferred to the server 30 (from (A) to (B) in FIG. 5. Switching).

  Thereafter, the server 30 takes over the processing of the server 20, and the provision of functions to the terminals connected to the network 50 is resumed.

  In (procedure c) described above, the server 20 stops providing functions. This is to prevent the internal state of the server 20 from changing due to the server 20 performing some processing while copying the state data from the server 20 to the server 30. For example, if the server 20 performs some processing while copying the state data from the server 20 to the server 30, inconsistency with the copied state data may occur in the server 30.

Japanese Patent No. 4112191

  In the operation of switching the server that provides the function, neither the operation server nor the backup server provides the function during the procedure c to the procedure f described above. For this reason, packets transmitted to the operation server during this period are not processed, and there is a problem that it is not possible to ensure high reliability in providing functions by the server.

  By devising the software processing on the server, the amount of status data copied from the operation server to the backup server can be reduced, the performance of the copy processing can be improved, etc. Efforts are being made to reduce the time it takes. However, it is difficult to make this time zero.

  In addition, when a failure occurs on the primary server, processing such as restarting the primary server software or backup server software is performed, but before the provision of functions is resumed, Packets sent to are not processed.

  It is an object of the present invention to provide a packet switch that makes it possible to ensure high reliability in providing functions by a server.

In order to achieve the above object, a packet switch according to the present invention includes a switch unit having a plurality of ports that input and output packets, a port that receives a packet among the plurality of ports, and a port that outputs the packet. A packet switch having a routing table stored in association with the destination address of
The switch unit receives port information indicating at least one of the plurality of ports and a first message including an address via any of the plurality of ports,
A buffer unit connected to the switch unit via a first port of the plurality of ports and receiving and storing a packet output from the first port;
When the first message is received, it is accepted by a port other than the port indicated by the port information included in the first message among the plurality of ports, and the address included in the first message is set as the destination address. And a control unit that sets the routing table so as to output the packet from the first port.

  Since the present invention is configured as described above, a packet transmitted to a server while the function is not provided by the server can be temporarily stored in the buffer unit.

  The packet temporarily stored in the buffer unit is transferred to the server that provides the function and processed after the provision of the function is resumed. Therefore, it is possible to ensure the high reliability of the function provision by the server.

It is a block diagram which shows the structure of one Embodiment of the packet switch of this invention. FIG. 2 is a block diagram illustrating an example of a configuration of a large-scale buffer unit illustrated in FIG. 1. FIG. 3 is a diagram for explaining an example of the operation of the packet switch shown in FIGS. 1 and 2. It is a figure for demonstrating the other example of operation | movement of the packet switch shown in FIG. 1 and FIG. It is a figure for demonstrating an example of an operation | movement when switching the server which provides a function.

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

  FIG. 1 is a block diagram showing a configuration of an embodiment of a packet switch of the present invention.

  As shown in FIG. 1, the packet switch 10 of the present embodiment includes a switch unit 11 having ports 11 a to 11 f for inputting and outputting packets, a control unit 12, a routing table unit 13, and a large-scale buffer unit 14. ing. In FIG. 1, the switch unit 11 has six ports, but the number is not limited to six.

  The routing table unit 13 stores a routing table. The routing table stores the port from which the packet is received among the ports 11a to 11f and the port that outputs the packet in association with each other according to the destination address of the packet. The address is, for example, an IP (Internet Protocol) address.

  The switch unit 11 transfers the packet according to the contents stored in the routing table. Specifically, when a packet is received at each of the ports 11a to 11f, the switch unit 11 outputs the packet by referring to the routing table from the received port and the destination address of the packet. To decide. Then, the switch unit 11 outputs the packet from the determined port. The switch unit 11 receives a message transmitted from a terminal or a server connected via the ports 11 a to 11 d and destined for the control unit 12. Details of this message will be described later.

  The control unit 12 controls the operation of the entire packet switch 10. Specifically, for example, setting and updating of the routing table, state management of the switch unit 11, and state management and control of the large-scale buffer unit 14 are performed. The control unit 12 is connected to the switch unit 11 via the port 11e, and receives a message received by the switch unit 11 via the port 11e. The message includes, for example, a stop request message that is a first message for requesting stop of packet transfer, and a stop release request message that is a second message for requesting restart of transfer of a stopped packet. The stop request message includes port information and an address indicating at least one of the ports 11a to 11f. When receiving the stop request message, the control unit 12 sets a routing table according to the port information and address included in the received stop request message. Further, when receiving the stop release request message, the control unit 12 determines a second port from the ports 11a to 11f based on the received stop release request message. And the control part 12 sets a routing table so that the packet received in the port 11f may be output from a 2nd port. The details of the operation of setting the routing table by the control unit 12 will be described in the operation flow described later.

  The large-scale buffer unit 14 is connected to the switch unit via the port 11f, which is the first port, and receives and stores a packet output from the port 11f.

  FIG. 2 is a block diagram showing an example of the configuration of the large-scale buffer unit 14 shown in FIG.

  As shown in FIG. 2, the large-scale buffer unit 14 shown in FIG. 1 includes a buffer memory 141, a reading unit 142, a writing unit 143, a management unit 144, a rate register 145, a timer 146, and a maximum storage time. A register 147 and a pattern table 148 are provided.

  The management unit 144 manages the state of the large-scale buffer unit 14. The control unit 12 sets a rate register 145, a maximum storage time register 147, and a pattern table 148, which will be described later, via the management unit 144. Further, the control unit 12 monitors the free state of the buffer memory 141 via the management unit 144.

  The writing unit 143 receives a packet output from the port 11 f and writes the received packet in an empty area of the buffer memory 141.

  The buffer memory 141 is a memory that stores packets. The size of the buffer memory 141 is determined by the product of the transfer rate of the ports 11a to 11f and the longest time for storing a packet. For example, when the transfer rate of the ports 11a to 11f is 1 Gbps and a packet is stored for a maximum of 80 seconds, the size of the buffer memory 141 is 10 Gbytes. In this case, the buffer memory 141 must have a performance capable of performing packet writing and packet reading in parallel at a speed of 1 Gbps.

  When the reading unit 142 receives an instruction to start reading from the control unit 12 via the management unit 144, the reading unit 142 reads and outputs the packet stored in the buffer memory 141.

  Here, the large-scale buffer unit 14 may include a rate register 145, a timer 146, a maximum storage time register 147, and a pattern table 148, as shown in FIG.

  In the rate register 145, a predetermined rate for outputting the packet stored in the buffer memory 141 is preset by the control unit 12. When the rate register 145 is provided, the reading unit 142 outputs a packet stored in the buffer memory 141 at a predetermined rate set in the rate register 145.

  In the pattern table 148, a predetermined pattern of the header portion of the packet is set in advance by the control unit 12. When the pattern table 148 is provided, when the writing unit 143 receives the packet output from the port 11f, the writing unit 143 determines whether the header portion of the received packet matches a predetermined pattern set in the pattern table 148. judge. As a result of this determination, when the header portion of the received packet matches a predetermined pattern set in the pattern table 148, the writing unit 143 writes the received packet in the buffer memory 141. On the other hand, when the header portion of the received packet does not match the predetermined pattern set in the pattern table 148, the writing unit 143 discards the received packet. In the pattern table 148, a plurality of patterns can be set.

  The timer 146 is a counter that indicates time. In the maximum storage time register 147, the maximum storage time, which is the maximum time for storing a packet, is preset by the control unit 12.

  When the timer 146 and the maximum storage time register 147 are provided, when the writing unit 143 receives the packet output from the port 11f, the writing unit 143 displays the time indicated by the timer 146 at the time of reception together with the received packet in the buffer memory 141. Write to an empty area. In this case, the size of the memory necessary for storing the received packet increases by the amount for storing the time. The reading unit 142 also reads the time when reading the packet stored in the buffer memory 141. Then, the reading unit 142 calculates a difference between the time indicated by the timer 146 when the packet is read and the time read from the buffer memory 141. When the calculated difference is smaller than the maximum storage time set in the maximum storage time register 147, the reading unit 142 outputs the read packet. On the other hand, when the calculated difference is equal to or longer than the maximum storage time set in the maximum storage time register 147, the reading unit 142 discards the read packet.

  Hereinafter, the operation of the packet switch 10 configured as described above will be described. Here, a case where the operation server software and hardware are systematically replaced will be described.

  FIG. 3 is a diagram for explaining an example of the operation of the packet switch 10 shown in FIGS. 1 and 2.

  As shown in FIG. 3, each of the network 50, the server 20, the server 30, and the server 40 is connected to the switch unit 11 via ports 11a, 11b, 11c, and 11d. Here, the server 20 is an operation server and the server 30 is a backup server.

  During normal operation, when a packet whose destination address indicates the server 20 is received at the port 11a, port 11c, and port 11d, the routing table is set so that the received packet is output from the port 11b. This is indicated by (1) and (5) in FIG.

  Further, during normal operation, no packet is stored in the buffer memory 141, and the reading unit 142 stops operating.

  (Procedure 1) The server 20 activates the server 30 while providing a function to a terminal or the like connected to the network 50 or the server 40.

  (Procedure 2) The server 20 copies various environmental data to the server 30 while continuing to provide functions.

  The operation so far is the same even when the packet switch 10 of this embodiment is not used.

  (Procedure 3) The server 20 stops providing functions to the terminal and the server 40 connected to the network 50. At this time, the server 20 transmits a stop request message including the port information indicating the port 11 c and the address of the server 20 to the control unit 12. When the control unit 12 receives the stop request message transmitted from the server 20 through the port 11e, the ports 11a, 11d other than the port 11c among the ports 11a, 11c, and 11d connected to the servers 30 and 40 and the network 50 are provided. The routing table is set so that a packet whose destination address indicates the server 20 is output from the port 11f. As a result, as shown in (2) of FIG. 3, a packet transmitted from the network 50 and the server 40 and whose destination address indicates the server 20 is transferred to the large-scale buffer unit 14. Note that a packet that is received at the port 11c indicated by the port information included in the received stop request message and whose destination address indicates the server 20 is continuously transferred according to (5) of FIG. That is, the packet transmitted from the server 30 is not transferred to the large-scale buffer unit 14. Therefore, communication can be performed between the server 20 and the server 30. Although not explicitly shown in FIG. 3, the communication between the ports 11 a to 11 f and the control unit 12 is managed by the control unit 12, and there is a stop request message transmitted from the server 20. Regardless, the transfer is not stopped.

  (Procedure 4) The server 20 copies the status data regarding the service being processed to the server 30.

  (Procedure 5) The server 20 shifts the server 30 to the operating state.

  The operations of Procedure 4 and Procedure 5 are the same even when the packet switch 10 of this embodiment is not used.

  (Procedure 6) A stop cancellation request message is transmitted to the control unit 12. At this time, there are a method in which the server 20 transmits a stop release request message including port information indicating the port 11c, and a method in which the server 30 transmits a stop release request message. In the former method, the second port, that is, a new transfer destination of the packet is explicitly specified. The former method is suitable not only when the operation server is instructed but also when an instruction to cancel the suspension of packet transfer is issued from, for example, a server that manages the operating state of the server group and server switching. On the other hand, in the latter method, the control unit 12 determines the second port, that is, the new transfer destination of the packet, by identifying the port from which the stop cancellation request message is accepted among the ports 11a to 11f. In the latter method, packet transfer can be resumed without setting port information indicating the port to which the backup server is connected to the backup server. When the control unit 12 receives the stop cancellation request message transmitted from the server 20 or the server 30 via the port 11e, the control unit 12 sets the routing table to output the packet received at the port 11f from the port 11c. This sets (3) shown in FIG. Then, the control unit 12 outputs a read start instruction to the large-scale buffer unit 14. As a result, the packet stored in the buffer memory 141 is output and accepted by the port 11f, and the accepted packet is outputted from the port 11c. That is, the packet stored in the buffer memory 141 is transferred to the server 30. At this point, packets that are accepted by the ports 11a and 11d and whose destination address indicates the server 20 are continuously output from the port 11f. This is because the newly accepted packet is transferred to the server 30 following the packet stored in the large-scale buffer unit 14. Thereby, reversal of the order of packets can be avoided. It should be noted that a packet received at the port 11b and whose destination address indicates the server 20 can be considered as a method of returning it as it is and outputting it from the port 11b, or a method of discarding it. Can be set by the user.

  According to the above procedure 6, the server 30 that has taken over the processing that the server 20 has performed as the operation server can provide the function.

  Next, in order to return the packet transfer via the large-scale buffer unit 14 to the state of packet transfer not via the large-scale buffer unit 14, the following procedure 7 is executed.

  (Procedure 7) The controller 12 monitors the availability of the buffer memory 141. Then, when there are no more packets stored in the buffer memory 141, the control unit 12 sets the routing table so that it is accepted by the port 11a and the port 11d and the packet whose destination address indicates the server 20 is output from the port 11c. . As a result, as shown in (4) of FIG. 3, a packet transmitted from the network 50 and the server 40 and whose destination address indicates the server 20 is transferred to the server 30. Then, the control unit 12 stops the operation of the reading unit 142. As a result, the large-scale buffer unit 14 is in a state that can cope with switching of another server.

  As described above, in the present embodiment, when the stop request message is received, the packet switch 10 is accepted by a port other than the port indicated by the port information included in the stop request message among the plurality of ports, and the stop request message The routing table is set so as to output from the port 11f a packet having the destination address as the destination address.

  Thereby, the packet transmitted to the operation server while the operation server and the backup server are not providing functions can be temporarily stored in the large-scale buffer unit 14.

  The packet temporarily stored in the large-scale buffer unit 14 is transferred to the backup server and processed after the provision of the function by the backup server is started. Therefore, it is possible to ensure the high reliability of the function provision by the server.

  In addition, the large-scale buffer unit 14 including the buffer memory 141 is connected to the switch unit 11 and can temporarily store packets accepted in any of the ports 11a to 11e. Therefore, the server can be switched by sharing the buffer memory 141 with a plurality of servers. Therefore, for example, compared to a case where the capacity of a buffer provided for each of a plurality of ports is expanded to temporarily store a packet, it is possible to economically ensure the high reliability of providing a function by the server. .

  In addition, a very large number of packets may be stored in the buffer memory 141 until the backup server completes preparation for providing the function. If a very large number of packets stored in the buffer memory 141 are output all at once immediately after the backup server starts providing the function, the backup server falls into a congestion state.

  This can be avoided by setting the predetermined rate set in the rate register 145 so that the backup server does not fall into a congestion state.

  Further, various types of packets may be transmitted from various terminals until the backup server completes preparation for providing the function. When the backup server starts providing the function, it may be considered not to process all of these packets, but to handle only those packets that are indispensable for continuing the function. It is done.

  By storing in the buffer memory 141 only the header portion of the packet that matches the predetermined pattern set in the pattern table 148, it becomes possible to transfer only the packet indispensable for continuing the provision of the function to the backup server.

  In addition, if a terminal or the like transmits a packet to the operation server before the backup server completes preparation for providing the function, the terminal or the like that transmitted the packet will not If there is no response to the packet, the packet may be retransmitted.

  When the terminal or the like performs such an operation, there may be a corresponding method in which, when the backup server starts providing a function, a packet that is old enough to be retransmitted by the terminal is not processed.

  By providing the timer 146 and the maximum storage time register 147, it is possible to avoid transferring packets that the backup server does not need to process to the backup server.

  Further, as described above, in the stop request message, it is not necessary to clearly indicate the transfer of the packet to the large-scale buffer unit 14. Therefore, there is no need to notify the operation server or the like of port information for identifying the port to which the large-scale buffer unit 14 is connected, or for the operation server or the like to manage the port information.

  The stop request message is not a message for instructing packet transfer to the large-scale buffer unit 14, but a message for instructing stop of packet transfer. Thereby, it is not necessary to change the format of the stop request message according to the presence or absence of the large-scale buffer unit 14. That is, when the packet switch receives the stop request message, the packet switch transfers the packet to the large-scale buffer unit 14 if the packet switch has the large-scale buffer unit 14, and if the packet switch does not have the large-scale buffer unit 14, the packet is transferred. It can be discarded.

  When the packet is transferred to the large-scale buffer unit 14 and there is no free area in the buffer memory 141, the packet stored at the earliest time among the plurality of packets stored in the buffer memory 141 is discarded. However, the newly transferred packet may be stored in the large-scale buffer unit 14. This is because, for a packet stored in the buffer memory 141 at an earlier time, there is a possibility that the terminal that transmitted the packet has started a retransmission procedure.

  Thus, when the backup server provides a function, a packet that is highly likely to be required for the function can be left in the large-scale buffer unit 14.

  In the above-described embodiment, the stop request message has been described as including the port information of the port to which the backup server is connected. However, in order to switch the server that provides the function, the operation server may communicate with an operation management server that manages the switching process. Therefore, the stop request message may include not only port information indicating the port to which the backup server is connected, but also port information of the port to which such an operation management server or the like is connected.

  Here, what has been described with reference to FIG. 3 is the operation of the packet switch 10 when the server that provides the function is switched from the operation server to the backup server when the operation server software or hardware is systematically replaced. Met.

  Next, the operation of the packet switch 10 when a failure or the like occurs in the operation server will be described.

  FIG. 4 is a diagram for explaining another example of the operation of the packet switch 10 shown in FIGS. 1 and 2. Here, the operation of the packet switch 10 when the operation server detects a software failure and the software in the operation server is restarted will be described.

  As shown in FIG. 4, each of the network 50, the server 20, and the server 40 is connected to the switch unit 11 via ports 11a, 11b, and 11d. Here, the server 20 is an operational server.

  During normal operation, when a packet whose destination address indicates the server 20 is received at the port 11a and the port 11d, the routing table is set so that the received packet is output from the port 11b. This state is indicated by (6) in FIG.

  Further, during normal operation, no packet is stored in the buffer memory 141, and the reading unit 142 stops operating.

  When the server 20 detects a software failure and starts restarting the software, a stop request message is transmitted to the control unit 12 at the start of restart. As a result, the control unit 12 sets the routing table so that the packet received at the ports 11a and 11d and the destination address indicating the server 20 is output from the port 11f. As a result, as shown in (7) of FIG. 4, a packet transmitted from the network 50 and the server 40 and having a destination address indicating the server 20 is transferred to the large-scale buffer unit 14.

  Next, when the server 20 completes the software restart, the server 20 transmits a stop cancellation request message to the control unit 12. As a result, the control unit 12 sets the routing table so that the packet received at the port 11f is output from the port 11b as shown in (8) of FIG. Then, the control unit 12 outputs a read start instruction to the large-scale buffer unit 14. Thereby, the packet transmitted to the server 20 while the software of the server 20 is restarted is processed after the software is restarted.

  Then, the control unit 12 monitors the free state of the buffer memory 141. When there are no more packets stored in the buffer memory 141, the control unit 12 accepts the packets at the ports 11a and 11d and the destination address indicates the server 20 from the port 11b as shown in (6) of FIG. Set the routing table to As a result, the normal operation state is restored.

  In the case of the operation described with reference to FIG. 4, a packet received by the server 20 between the occurrence of a failure in the operation server and the detection of the failure is not processed normally, but after the failure is detected. Are normally processed.

  Therefore, when viewed from a terminal or the like connected to the network 50, the apparent interruption time is greatly reduced as compared with the case where the packet switch 10 is not used. This is the same not only when the software is restarted on the operation server, but also when the data in the operation server is reinitialized or the redundant device in the operation server is switched. .

  That is, even when a failure or the like occurs in the operation server, it is possible to ensure the high reliability of the function provision by the server.

  The present invention can be easily realized by adding the large-scale buffer unit 14 to the conventional packet switch and changing the software of the control unit 12, and can be applied to various packet switches.

  In particular, it is highly possible to accommodate a large number of servers that realize continuity of function provision while handing over state data between servers, and to the large number of users due to interruption of server function provision. It is assumed that the present invention is applied to a packet switch or the like constituting a large-scale data center that is highly likely to be affected.

DESCRIPTION OF SYMBOLS 10 Packet switch 11 Switch part 11a-11f Port 12 Control part 13 Routing table part 14 Large scale buffer part 20, 30, 40 Server 50 Network 141 Buffer memory 142 Reading part 143 Writing part 144 Management part 145 Rate register 146 Timer 147 Maximum storage Time register 148 Pattern table

Claims (9)

A switching unit having a plurality of ports for inputting and outputting packets; a routing table for storing a port that receives a packet and a port that outputs the packet among the plurality of ports in association with each other according to a destination address of the packet; A packet switch comprising:
The switch unit receives port information indicating at least one of the plurality of ports and a first message including an address via any of the plurality of ports,
A buffer unit connected to the switch unit via a first port of the plurality of ports and receiving and storing a packet output from the first port;
When the first message is received, it is accepted by a port other than the port indicated by the port information included in the first message among the plurality of ports, and the address included in the first message is set as the destination address. And a control unit that sets the routing table so as to output the packet from the first port. The packet switch according to claim 1,
The switch unit receives the second message via any of the plurality of ports;
When the second message is received, the control unit determines a second port from the plurality of ports based on the received second message, and accepts the second port from the first port. A packet switch configured to output the packet stored in the buffer unit after setting the routing table to output the received packet from the second port. The packet switch according to claim 2, wherein
The second message includes port information indicating one of the plurality of ports,
The control unit is a packet switch in which a port indicated by port information included in the received second message is the second port. The packet switch according to claim 2, wherein
The control unit is a packet switch in which the port from which the second message is received is the second port among the plurality of ports. The packet switch according to any one of claims 2 to 4, wherein
When the packet stored in the buffer unit is exhausted, the control unit is accepted by a port other than the port indicated by the port information included in the first message among the plurality of ports, and the first message A packet switch that sets the routing table so that a packet having a destination address as an included address is output from the second port. The packet switch according to any one of claims 2 to 5,
The buffer unit is a packet switch that outputs the stored packets at a predetermined rate. The packet switch according to any one of claims 2 to 6,
When the buffer unit outputs the stored packet, the buffer unit determines whether the time in which the packet is stored is shorter than a predetermined time, and the time in which the packet is stored is shorter than the predetermined time. A packet switch that outputs the packet when it is short. The packet switch according to any one of claims 1 to 7,
The buffer unit is a packet switch that stores a received packet when a header portion of the received packet matches a predetermined pattern. The packet switch according to any one of claims 1 to 8,
When there is no empty area for storing the received packet, the buffer unit discards the packet stored at the earliest time among the plurality of packets stored in the buffer unit and stores the received packet. Packet switch to do.

Images