JP2014003700A - Relay device and communication system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique for detecting failure occurring in a relay device.SOLUTION: A first relay device comprises: first and second network interfaces; a transfer unit which relays packets received via the first network interface to a second network via the second network interface by using a predetermined transmission band; a band control unit which controls the predetermined transmission band on the basis of a communication state between a second relay device and the first relay device; and a packet processing unit which processes data of a predetermined area in the relayed packets when the band control unit controls the predetermined transmission band.

Description

  The present invention relates to a relay device arranged between terminals for relaying data, a control method for the relay device, and a network system including the terminal and the relay device.

  As a communication network between global bases, it is common to use a WAN (Wide Area Network) using IP-VPN technology or the like. When a terminal at a base communicates with a terminal at another overseas base, the bases are connected via a WAN for communication. In the WAN, the available bandwidth is usually limited by the contract. In addition, the delay is particularly large in communications with overseas.

  TCP is generally used for communication between terminals. In TCP, the communication band is controlled according to the size of RTT (Round Trip Time). In communication via WAN, it may be difficult to obtain high throughput. Here, as a technique for realizing high throughput, there is a technique described in Patent Document 1.

International Publication Number WO2011 / 033894

  In communication using TCP, the transmission bandwidth is greatly affected by the RTT and the discard rate, so in an environment such as WAN where the RTT is large, the number of hops is large, and there are many discard locations, the transmission bandwidth is significantly lower than the contract bandwidth. There are cases where it can only be obtained.

  As countermeasures against this, by managing the stream so that the transmission bandwidth is not affected by the RTT and the discard rate using a plurality of relay devices in the middle of the transmission path of the stream type communication using the sequence number such as TCP. It is conceivable to improve the transmission band.

  At this time, if a failure occurs in the relay device, the failure cannot be detected by the relay device that faces the relay device in which the failure has occurred, and various problems may occur. For example, a problem that it takes time until communication between communication devices (terminals) is disconnected, a problem that a session remains in the relay device and communication on the same port is obstructed, or a memory shortage occurs. Etc. may occur.

  The present invention has been made to solve the above-described problem, and provides a technique capable of detecting the occurrence of a failure of an opposing relay device in an environment where stream-type communication using a sequence number is performed. With the goal.

  SUMMARY An advantage of some aspects of the invention is to solve at least a part of the problems described above, and the invention can be implemented as the following forms or application examples.

Application Example 1 A first terminal for relaying data between two terminals that are arranged in series on a transmission path of the two terminals and two terminals that transmit data by stream-type communication using a sequence number And a second relay device, the first relay device arranged in a network,
A transfer control unit for transferring received data;
A detection unit that detects that a failure has occurred in the second relay device that faces the second relay device;
The transfer control unit
A transfer unit for transferring the received data;
A processing unit that processes the received data to detect occurrence of a failure in the second relay device, and (i) data to be transferred when data is transferred to the second relay device And (ii) a transmission source when transmitting data to the opposite terminal, (ii) an option unit for adding to the data a mark for indicating data exchange between the first and second relay devices. An adder for adding a predetermined value to a sequence number assigned to data transmitted from the terminal, and (iii) a confirmation response to the data transmitted from the terminal when data is received from the opposite terminal A subtraction unit that subtracts the predetermined value from the confirmation response number when a number is given, and a processing unit,
The detection unit is a first relay device that detects occurrence of a failure in the second relay device facing the mark based on at least one of the mark and the confirmation response number.

  According to the first relay device described in Application Example 1, the processing unit can easily cause a failure of the second relay device facing the detection unit based on at least one of the mark and the confirmation response number. Can be detected.

[Application Example 2] The first relay device according to Application Example 1,
The detection unit is a first relay device that detects occurrence of a failure in the second relay device when the mark is not attached to the data transmitted and received from the second relay device.
According to the first relay device described in Application Example 2, it is possible to easily detect the occurrence of a failure in the second relay device by determining whether or not a mark is attached to the received data.

[Application Example 3] The first relay device according to Application Example 1,
The detection unit, when the confirmation response number of the data transmitted and received from the second relay device is a number obtained by adding the predetermined value to a number that should be received originally, A first relay device that detects the occurrence of a device failure.
According to the first relay device described in Application Example 3, by determining whether the confirmation response number in the received data is a number obtained by adding the predetermined value from a number that should be received originally. The occurrence of a failure in the second relay device can be easily detected.

[Application Example 4] The first relay device according to Application Example 1,
The detection unit is a case where the mark is not attached to the data transmitted and received from the second relay device, and the confirmation response number of the received data is a number that should be received originally A first relay device that detects the occurrence of a failure in the second relay device when the number is a number obtained by adding the predetermined value to the first relay device.
According to the first relay device described in the application example 4, the accuracy of the failure occurrence detection can be improved by performing the failure detection using the two elements of the mark and the confirmation response number.

[Application Example 5] The first relay device according to any one of Application Examples 1 to 4,
The two terminals communicate using TCP,
The transfer control unit
If the received data includes a SYN packet for making a TCP connection establishment request when the detection unit detects a failure of the second relay device based on the received data, the function of the processing unit is A first relay device that transfers the received data without using it.
According to the first relay device described in Application Example 5, data communication can be continued even when a failure is detected.

[Application Example 6] The first relay device according to any one of Application Examples 1 to 5,
The two terminals communicate using TCP,
The transfer control unit
When the detection unit detects the occurrence of a failure of the second relay device based on the received data, if the received data includes an acknowledgment packet for a SYN packet for making a TCP connection establishment request, the processing A first relay device that transfers the received data without using the function of the unit.
According to the first relay device described in Application Example 6, data communication can be continued even when a failure is detected.

  Note that the present invention can be realized in various modes. For example, not only the first relay device described above, but also a control method of the first relay device, a function of the first control device, a computer program for realizing the control method, and a storage storing the computer program It can be realized in a mode such as a medium, a first and second relay device, and a network system including the first and second communication devices.

It is a figure for demonstrating the network system 5w of a reference example. It is a figure for demonstrating the network system 5 as an Example. It is a figure which shows the TCP header used in an Example. It is a figure for demonstrating the change of the TCP head at the time of data communication in the network system. It is a processing flow of a three-way handshake that is normally performed. It is a figure for demonstrating the communication after connection establishment. It is a figure for demonstrating the detection of the failure generation | occurrence | production at the time of 3 way handshake. It is a figure for demonstrating the 1st aspect of the failure detection at the time of 3 way handshake. It is a figure for demonstrating the 2nd aspect of the failure detection at the time of 3 way handshake. It is a figure for demonstrating the 3rd aspect of the failure detection at the time of 3 way handshake. It is a figure for demonstrating the 4th aspect of the failure detection at the time of 3 way handshake. It is a figure for demonstrating the 5th aspect of the failure detection at the time of 3 way handshake. It is a figure for demonstrating the 6th aspect of the failure detection at the time of 3 way handshake. It is a figure for demonstrating the 1st failure detection aspect when a failure generate | occur | produces during the data communication by TCP. It is a process for demonstrating the 2nd failure detection aspect when a failure generate | occur | produces during the data communication by TCP.

Next, embodiments of the present invention will be described in the following order.
A. Reference example:
B. Example:
C. Variations:

A. Reference example:
Before describing the embodiment, a reference example to which the present invention is not applied will be described for easy understanding. FIG. 1 is a diagram for explaining a network system 5w of a reference example. The network system 5w includes a first communication device 20w as a terminal, a second communication device 21w as a terminal, a first relay device 10w, and a second relay device 11w. The first and second communication devices 20w and 21w realize data communication with the first and second relay devices 10w and 11w via the Internet.

  In the network system 5w, data communication between the first communication device 20w and the first relay device 10w and data communication between the second communication device 21w and the second relay device are managed by a TCP session. In the network system 5w, data communication between the first relay device 10w and the second relay device 11w is performed by a unique TCP session between the relay devices 10w and 11w (“high-speed TCP session”). It is also managed by the relay devices 10w and 11w.

  In the high-speed TCP session between the first and second relay apparatuses 10w and 11w, a sequence number ("original sequence number") assigned to data transmitted from the first communication apparatus 20w or the second communication apparatus 21w A sequence number (also referred to as “inter-relay device sequence number”) obtained by adding a certain value is used. For example, the second relay device 11w adds a certain value to the sequence number given to the data transmitted from the second communication device 21w. Then, the added sequence number is assigned to the data, and the data is transmitted to the first relay device 10w. When data is transferred from the relay devices 10w and 11w to the communication devices 20w and 21w, the value added from the inter-relay device sequence number is subtracted to return to the original sequence number. Further, in the high-speed TCP session between the first and second relay apparatuses 10w and 11w, a TCP option for identifying that the communication is between the relay apparatuses 10w and 11w is given to the data. The TCP option is deleted from the transmitted data when the data is transmitted from the relay devices 10w and 11w to the communication devices 20w and 21w.

  Here, consider a case where a failure occurs in the second relay device 11w (when a failure occurs). Here, “occurrence of failure” refers to a case where the function of transferring data (packets) is normal, but an abnormality has occurred in the function of transferring data using an accelerated TCP session. In other words, “occurrence of failure” means that the second relay apparatus 11w cannot add a certain value to the original sequence number, or the second relay apparatus 11w cannot add or delete the TCP option data. State.

  When a failure occurs, the first relay device 10w adds a certain value to the sequence number of the data transmitted from the first communication device 20w, and the data with the inter-relay device sequence number assigned to the second sequence number. Transmit to the relay device 11w. The second relay device 11w transmits the data to the second communication device 21w without processing the inter-relay device sequence number assigned to the received data. Since the sequence number of the received data is different from the sequence number (original sequence number) in the TCP session, the second communication device 21w recognizes the received data as illegal data. Then, the second communication device 21w transmits a sequence number (original sequence number) used in the TCP session and ACK. However, the first relay device 10w cannot recognize the sequence number transmitted from the second communication device 21w and cannot transfer data to the first communication device 20w. That is, there arises a problem that communication between the first and second communication devices 20w and 21w cannot be continued and the session cannot be disconnected.

B. Example:
B-1. Configuration of network system 5:
FIG. 2 is a diagram for explaining a network system 5 as an embodiment of the present invention. As shown in FIG. 2, the network system 5 includes first and second communication devices 454 and 455 as terminals, and first and second communication devices that relay data communication between the first and second communication devices 454 and 455. 2 relay devices 441 and 442. The first and second relay devices 441 and 442 are arranged in series on the transmission path of the first and second communication devices 454 and 455. The first and second communication devices 454 and 455 perform data (packet) communication using TCP. Note that the Internet (not shown) exists between the first and second relay apparatuses 441 and 442.

  The first relay device 441 includes input / output interfaces 41 and 42 for receiving and transmitting data with an external device (for example, the first communication device 454), and transfer control for controlling transfer of the received data. Unit 44 and a detection unit 46 for detecting a failure of another relay device 442. Note that the second relay device 442 has the same configuration, and thus the illustration of the internal configuration is omitted and the description thereof is omitted.

  The transfer control unit 44 includes a processing unit 48 for processing data to be transmitted and received, and a transfer unit 49 for transferring data. The transfer unit 49 transfers data based on an IP address or the like given to the received data.

  The processing unit 48 includes an option unit 48a, an addition unit 48b, and a subtraction unit 48c. When the option unit 48a transfers data toward the second relay device 442 facing the option unit 48a, the option unit 48a adds a mark to the transferred data to indicate that the data is exchanged between the first and second relay devices. Give to data. In this embodiment, this mark uses the TCP option of the TCP header. When the adder 48b transmits data to the opposite communication device (here, the first communication device 454), the adder 48b converts the data transmitted from the transmission source communication device (here, the second communication device 455) into data. A predetermined value is added to the assigned sequence number (original sequence number). An arbitrary value can be selected as the predetermined value, and in this embodiment, 1000000 is added to the original sequence number. When the subtraction unit 48c receives data from the opposite communication device (here, the first communication device 454), the subtraction unit 48c sets a predetermined value (this embodiment) to the confirmation response number assigned to the data transmitted from the communication device. In the example, 1000000) is subtracted. The first and second relay devices 441 and 442 include a session management table (not shown) for managing unique communication between the relay devices 441 and 442.

  The processing unit 48 may have a function for performing high-speed data communication between the first and second relay apparatuses 441 and 442. For example, the content of the disclosure may be incorporated in International Publication Number WO2011 / 033894 (International Application Date: August 19, 2010). For example, the processing unit 48 of the first and second relay apparatuses 441 and 442 may include a band control unit that controls the transmission band in order to transmit data based on the data retransmission status. When the bandwidth control unit is provided and the processing unit 48 functions normally, the transfer unit 49 transfers data according to the controlled transmission band.

  The detection unit 46 detects the occurrence of a failure in the opposing second relay device 442 based on at least one of the TCP option and the confirmation response number. Further, for example, the occurrence of a failure detected by the detection unit 46 is transmitted as failure occurrence information to a server (not shown). Then, the server performs processing such as analysis of failure occurrence information and notification to the outside.

  FIG. 3 is a diagram illustrating a TCP header used in the embodiment. The TCP header shown in FIG. 3 has a normal header configuration defined by TCP. The field 101 stores a source port number and a destination port number. A field 102 stores a sequence number. The field 103 stores an acknowledgment number (ACK number). The field 104 stores information such as a header length, various flags such as an ACK flag, and a window size. The field 105 stores a checksum and an urgent pointer. The field 106 stores an option type, an option length, and an option value. Specifically, the field 106 stores information on MSS (Maximum Segment Size). The field 107 stores an option type, an option length, and an option value (version number). Specifically, information on the TCP option assigned to the data at the time of data communication between the first and second relay apparatuses 441 and 442 is stored. The field 108 is a data field.

  FIG. 4 is a diagram for explaining a change in the TCP head during data communication in the network system 5. FIG. 4 shows an example of a part of communication performed at the time of connection establishment (at the time of 3-way handshake). The first communication device 454 starts communication with normal TCP executed between terminals (step S30). For example, a SYN packet is transmitted with a sequence number (Seq #) 10,000. The option unit 48a of the first relay device 441 adds a TCP option to the received data, and the transfer unit 49 transmits the data to the second relay device 442 (step S32). The option unit 48a of the second relay device 442 removes the TCP option from the received data, and the addition unit 48b of the second relay device adds 1000000 to the sequence number of the received data (step S33). Then, the transfer unit 49 transmits the processed data to the second communication device 455 (step S33).

  The second communication device 455 performs normal TCP processing, and adds the sequence expected next to the sequence number in the received data to the ACK number and transmits it (step S34). In this embodiment, the ACK number is a value obtained by adding the predetermined values 1000000 and 1 to the original sequence number transmitted in step S30. The subtraction unit 48c of the second relay apparatus 442 subtracts a predetermined value of 1000000 from the ACK number of the data transmitted from the transmission source second communication apparatus 455 so that the original sequence can be matched. Process (step S35). Also, the option unit 48a of the second relay device 442 gives a TCP option to the data (step S35). Then, the transfer unit 49 of the second relay device 442 transmits the processed data to the first relay device 441 (step S35). The option unit 48a of the first relay device 441 removes the TCP option of the data received from the second relay device 442 and matches it with the original stream (step S36). Also, the first relay device 441 adds 1000000, which is a predetermined value, to the sequence number of the received data (step S36). Then, the transfer unit 49 of the first relay device 441 transmits the processed data to the first communication device 454 (step S36). Although illustration is omitted, after step S36, the first relay device 441 transmits an ACK packet to the first relay device 441 using the ACK number of the received SYN / ACK packet as a sequence number. Then, the processing units 48 of the first relay device 441 and the second relay device 442 each perform predetermined processing on the ACK packet. Eventually, the second communication device 455 receives the ACK packet and a connection is established.

  In order to facilitate understanding, a three-way handshake process that is generally performed will be described. FIG. 5 is a processing flow of a three-way handshake that is normally performed. That is, FIG. 5 is a general processing flow at the time of three-way handshake when the present invention is not applied. The first communication device 454 transmits a SYN packet with an arbitrary sequence number A (an arbitrary number is input to A) to the second communication device 455 (step S40). As a response to the SYN packet, the second communication device 455 adds 1 to the received sequence number A by using an arbitrary sequence number B (an arbitrary number different from A is input to B). The number is set to the ACK number and the SYN / ACK packet is transmitted to the first communication device 454 (step S41). The first communication device 454 uses a value obtained by adding 1 to the sequence number A used in the SYN packet for the received SYN / ACK packet (that is, the ACK number transmitted in step S41) as a sequence number ( Step S42). Further, the first communication device 454 sets a value (B + 1) obtained by adding 1 to the sequence number B transmitted in step S41 as an ACK number (step S42). And the data which provided the sequence number and the ACK number are transmitted to the 2nd communication apparatus 455 (step S42). As a result, a connection is established between the first and second communication devices 454 and 455.

  FIG. 6 is a diagram for explaining communication after the connection of the network system 5 is established. In FIG. 6, it is assumed that the amount of data transmitted from the first communication device 454 is 1460 bytes. The first communication device 454 attaches a predetermined sequence number (sequence number 2001 in this embodiment) and transmits data to the first relay device 441 (step S50). The first relay device 441 adds a TCP option to the received data and transmits the data to the second relay device 442 (step S51). By giving the TCP option to the data, it is possible to distinguish from communication using normal TCP. When receiving the data transmitted from the first relay device 442, the second relay device 442 removes the TCP option from the received data and adds a predetermined value (1000000 in this embodiment) to the sequence number (step 1). S52). Then, the second relay device 442 transmits the processed data to the second communication device 455 (step S52).

  The second communication device 455 transmits data to which the ACK number in a state where a predetermined value (1000000 in this embodiment) is added to the number expected to be received next is added to the second relay device 442 (step S53). ). When the second relay device 442 receives the data transmitted from the second communication device 455, the second relay device 442 adds a TCP option to the received data and subtracts a predetermined value (1000000 in this embodiment) from the confirmation response number. (Step S54). Thereby, it becomes an ACK number with respect to the original sequence number transmitted at step S50. Then, the second relay device 442 transmits the processed data to the first relay device 441 (step S54). When receiving the data transmitted from the second relay device 442, the first relay device 441 removes the TCP option from the received data and adds a predetermined value (1000000 in this embodiment) to the sequence number (step 1). S55). Then, the first relay device 441 transmits the processed data to the first communication device 454 (step S55).

  The network system 5 performs the same processing for data communication (steps S56 to S58) after step S55. That is, the operation of step S56 is the same as the operation of step S50, the operation of step S57 is the same as the operation of step S51, and the operation of step S58 is the same as the operation of step S52.

  In the following, a method for detecting a failure occurrence in each scene in the first and second relay apparatuses 441 and 442 of the network system 5 will be described in detail.

B-2. About detection of failure occurrence:
Detection of failure in B-2-1.3 way handshake:
FIG. 7 is a diagram for explaining detection of the occurrence of a failure during a three-way handshake. That is, FIG. 7 shows that a failure occurs when a failure (specifically, a failure of the processing unit 48) occurs in either the first relay device 441 or the second relay device 442 during the three-way handshake. It is a figure for demonstrating the detection method of.

In TCP, when establishing a connection, a SYN / ACK packet is returned to the SYN packet, and a connection is established by returning an ACK packet to the SYN / ACK packet. Specifically, the connection is established by the steps described below.
Step S60: A SYN packet is transmitted from the first communication device 454 to the first relay device 441.
Step S61: The data is transmitted from the first relay device 441 to the second relay device 442 with the TCP option added to the data.
Step S62: The TCP option is removed from the data, and a predetermined value of 1000000 is added to the sequence number, and the data is transmitted from the second relay device 442 to the second communication device 455.
Step S63: SYN / ACK is transmitted from the second communication device 455 to the second relay device 442. For the ACK number, 1 is added to the received sequence number.
Step S64: A TCP option is assigned to the data, and a predetermined value of 1000000 is subtracted from the ACK number, and the data is transmitted from the second relay device 442 to the first relay device 441.
Step S65: The TCP option is removed from the data, and a predetermined value of 1000000 is added to the sequence number, and the data is transmitted from the first relay device 441 to the first communication device 454.
Step S66: The received ACK number is set as a sequence number, 1 is added to the received sequence number, and ACK is transmitted from the first communication device 454 to the first relay device 441.
Step S67: The TCP option is added to the data, and the data is transmitted from the first relay device 441 to the second relay device 442.
Step S68: The TCP option is removed from the data, and a predetermined value of 1000000 is added to the sequence number, and the data is transmitted from the second relay device 442 to the second communication device 455.

  Here, the outline of the failure detection method when a failure occurs at each of the points 60p, 61p, 64p, 65p, 66p, and 67p shown in FIG. 7 is as follows.

-When a failure occurs at point 60p (first failure detection mode):
Since the TCP option is not given to the SYN packet, the second relay device 442 can detect a failure.
・ When a fault occurs at point 61p (second fault detection mode)
Since the TCP option is not assigned to the SYN / ACK packet, the first relay device 441 can detect a failure.
・ When a failure occurs at point 64p (third failure detection mode)
Since the TCP option is not added to the SYN / ACK packet and the ACK number is not subtracted, the first relay device 441 can detect a failure.
・ When a failure occurs at point 65p (fourth failure detection mode)
Since the TCP option is not added to the ACK packet, the second relay device 442 can detect a failure.
・ When a fault occurs at point 66p (fifth fault detection mode)
Since the TCP option is not added to the ACK packet and the ACK number is not subtracted, the second relay device 442 can detect a failure.
・ When a fault occurs at point 67p (sixth fault detection mode)
Since the TCP option is not added to the retransmitted SYN / ACK packet and the ACK number is not subtracted, the first relay device 441 can detect a failure.

  A detailed processing flow of the occurrence of a failure at each of the points 60p, 61p, 64p, 65p, 66p, and 67p listed above will be described in detail below as an embodiment of failure detection.

B-2-2. First failure detection mode:
FIG. 8 is a diagram for explaining a first mode of failure detection during a three-way handshake. FIG. 8 is a processing flow in the case where a failure has occurred in the first relay device 441 at the time point 60p when the first communication device 454 transmits a SYN packet.

  When the first communication device 454 transmits a SYN packet, the first relay device 441 cannot process the SYN packet by the processing unit 48, and passes through the SYN packet (step S70). The detection unit 46 of the second relay device 442 can detect the occurrence of a failure in the first relay device 441 by detecting that the TCP option is not added to the received SYN packet (step S70a). Steps subsequent to step S71 are unique sessions (accelerated speed) that can be used between the first and second relay apparatuses 441 and 442 between the first and second relay apparatuses 441 and 442 of the network system 5. Since no session is formed, data is transferred by a normal TCP session. That is, after step S71, the second communication device 455 transmits a SYN / ACK packet to the second relay device 442, and the second relay device 442 and the first relay device 441 transmit the SYN / ACK packet. It transfers toward the 1st communication apparatus 454 (step S73, S73a). The first communication apparatus 454 that has received the SYN / ACK packet adds a value of 1 to the sequence number 10000 used in the SYN packet to the received SYN / ACK packet (that is, the ACK number transmitted in step S72). ) Is transmitted as a sequence number to the second communication device 455 (step S74). The first relay device 441 and the second relay device 442 pass through the ACK packet transmitted from the first communication device 454 and forward it to the second communication device 455 (steps S75 and S76). As a result, a connection is established between the first communication device 454 and the second communication device 455.

B-2-3. Second failure detection mode:
FIG. 9 is a diagram for explaining a second mode of failure detection during a three-way handshake. FIG. 9 is a processing flow when a failure has occurred in the second relay device 442 at the time 61p when the SYN packet is transmitted from the first communication device 454.

  In order to establish a connection, a SYN packet is transmitted from the first communication device 454 to the first relay device 441 (step S80). When receiving the SYN packet from the first communication device 454, the first relay device 441 adds a TCP option to the SYN packet and transmits it to the second relay device 442 (step S81). However, since the second relay device 442 cannot perform predetermined processing in the processing unit 48 (FIG. 2), the second relay device 442 passes through the SYN packet and transmits it to the second communication device 455 (step S81a). The second communication device 455 adds the TCP option to the received SYN packet. However, since the second communication device 455 does not support the TCP option and cannot interpret it, the TCP option is ignored (step S81b). . The second communication device 455 transmits a SYN / ACK packet to the second relay device 442 as a response to the received SYN packet (step S82). Here, the second relay device 442 transmits the SYN / ACK packet transmitted from the second communication device 455 to the first relay device 441 without being processed by the processing unit 48 (step S82a). The first relay device 441 can detect the occurrence of a failure in the second relay device 442 by detecting that the TCP option is not added to the SYN / ACK packet received from the second relay device 442 (step S40). S82b). In step S82b, the first relay device 441 deletes the session management table after detecting the occurrence of a failure, and passes through the subsequent communication.

  The first relay device 441 transmits the SYN / ACK packet to the first communication device 454 without using the function of the processing unit 48 in step S83 (pass-through). The first communication device 454 uses the value obtained by adding 1 to the sequence number 10000 used in the SYN packet to the received SYN / ACK packet (that is, the ACK number transmitted in step S83) as an ACK packet. Is transmitted to the second communication device 455 (step S84). In step S84, the ACK packet is sequentially transferred to the first relay device 441, the second relay device 442, and the second communication device 455 by pass-through through a normal TCP session (steps S85 and S86). As a result, a connection is established between the first communication device 454 and the second communication device 455.

B-2-4. Third failure detection mode:
FIG. 10 is a diagram for explaining a third mode of failure detection during a three-way handshake. FIG. 10 shows a processing flow when a failure occurs in the second relay device 442 at the point 64p where the second relay device 442 transmits the SYN packet to the second communication device 455.

  In FIG. 10, steps S90 to S92 are the same processing steps as steps S60 to S62 shown in FIG. After step S92, the second communication device 455 transmits a SYN / ACK packet to the second relay device 442 as a response to the SYN packet (step S93). Here, the second relay device 442 passes through the received SYN / ACK packet and transmits the SYN / ACK packet to the first relay device 441 (step S93a). The first relay device 441 detects that the TCP option is not added to the received SYN / ACK packet and that the ACK number is not subtracted, so that the failure of the opposing second relay device 442 occurs. Can be detected (step S94). In this embodiment, the fact that the ACK number is not subtracted means that the value obtained by subtracting the predetermined value (1000000) from the received ACK number by the detection unit 46 of the second relay apparatus 442 is expected to be received. Judged by the number. After detecting the failure of the opposing second relay device 442, the first relay device 441 is reset to the second communication device 455 via the first communication device 454 and the second relay device 442. A (RST) packet may be transmitted, and subsequent communication may be forcibly terminated. In addition, after detecting the failure of the opposing second relay device 442, the first relay device 441 passes through the received SYN / ACK packet and transmits the SYN / ACK packet to the first communication device 454. The three-way handshake process may be continued.

B-2-5. Fourth failure detection mode:
FIG. 11 is a diagram for explaining a fourth mode of failure detection during a three-way handshake. FIG. 11 is a processing flow when a failure occurs in the first relay device 441 at a point 65p where the first relay device 441 transmits the SYN packet to the second relay device 442.

  In FIG. 11, steps S100 to S103 are the same processing steps as steps S60 to S63 shown in FIG. For the SYN / ACK packet received from the second communication device 455, the second relay device 442 adds a TCP option and subtracts a predetermined value from the acknowledgment number to send the SYN / ACK packet to the first relay device. It transmits to 441 (step S104). Since the first relay device 441 has a fault in the processing unit 48, the first communication device 454 does not process the SYN / ACK packet received from the second relay device 442 by the processing unit 48. (Step S104a). Although the TCP option remains attached to the SYN / ACK packet transmitted to the first communication device 454, since the first communication device 454 cannot interpret the TCP option, the SYN / ACK packet to which the TCP option is attached. Can be received (step S104b). The first communication device 454 transmits an ACK packet to the first relay device 441 using the ACK number of the received SYN / ACK packet as a sequence number (step S105). The first relay apparatus 441 passes through the ACK packet transmitted from the first communication apparatus 454 and transmits it to the second relay apparatus 442 (step S105a). The second relay device 442 can detect the occurrence of a failure in the opposing first relay device 441 by detecting that the TCP option is not added to the ACK packet transmitted from the first relay device 441 ( Step S106). After detecting the failure of the opposing first relay device 441, the second relay device 442 communicates with the first communication device 454 via the second communication device 455 and the first relay device 441. An RST packet may be transmitted to forcibly end subsequent communication. Further, after detecting the failure of the opposing first relay device 441, the second relay device 442 passes through the received ACK packet and transmits the ACK packet to the second communication device 455. The 3-way handshake process may be continued.

B-2-6. Fifth fault detection mode:
FIG. 12 is a diagram for explaining a fifth mode of failure detection during a three-way handshake. FIG. 12 is a processing flow when a failure occurs in the first relay device 441 at the point 66p where the first relay device 441 transmits the SYN / ACK packet to the first communication device 454.

  In FIG. 12, steps S110 to S115 are the same processing steps as steps S60 to S65 shown in FIG. After step S115, the first communication device 454 transmits an ACK packet to the first relay device 441 using the ACK number of the received SYN / ACK packet as a sequence number (step S116). Since the failure has occurred in the processing unit 48, the first relay device 441 transmits the received ACK packet to the second relay device 442 without being processed by the processing unit 48 (step S117). The second relay device 442 detects the failure of the opposing first relay device 441 by detecting that the TCP option is not added to the received ACK packet and that the ACK number is not subtracted. Yes (step S118). After detecting the failure of the opposing first relay device 441, the second relay device 442 is reset to the first communication device 454 via the second communication device 455 and the first relay device 441. A (RST) packet may be transmitted, and subsequent communication may be forcibly terminated. In addition, after detecting the failure of the opposing first relay device 441, the second relay device 442 passes through the received SYN / ACK packet and transmits the SYN / ACK packet to the first communication device 454. The three-way handshake process may be continued.

B-2-7. Sixth failure detection mode:
FIG. 13 is a diagram for explaining a sixth mode of fault detection during a three-way handshake. FIG. 13 is a processing flow when a failure occurs in the second relay device 442 immediately after the second relay device 442 transmits a SYN / ACK packet to the first relay device 441.

  In FIG. 13, steps S120 to S126 are the same processing steps as steps S60 to S66 shown in FIG. The first relay device 441 adds a TCP option to the ACK packet transmitted from the first communication device 454, and subtracts a predetermined value of 1000000 from the ACK number to send the ACK packet to the second relay device 442. (Step S127). Since the processing unit 48 has a failure, the second relay device 442 transmits the ACK packet received from the first relay device 441 to the second communication device 455 without being processed by the processing unit 48 ( Step S128). The second communication device 455 receives the ACK packet, but the sequence number of the received ACK packet is different from the expected sequence number (sequence number: 1000011). 2 is retransmitted to the second relay device 442 (step S129). Also, the second relay device 442 passes through the retransmitted SYN / ACK packet and transmits it to the first relay device 441 (step S129a). The first relay device 441 detects that the TCP option is not added to the received SYN / ACK packet and that the ACK number is not subtracted, so that the failure of the opposing second relay device 442 occurs. Can be detected (step S130). After detecting the failure of the opposing second relay device, the first relay device 441 resets to the second communication device 455 via the first communication device 454 and the second relay device 442 ( RST) packet may be transmitted, and subsequent communication may be forcibly terminated. In addition, after detecting the failure of the opposing second relay device 442, the first relay device 441 passes through the received SYN / ACK packet and transmits the SYN / ACK packet to the first communication device 454. The three-way handshake process may be continued.

B-3. About detection of failure after connection is established:
FIG. 14 is a diagram for explaining a first failure detection mode when a failure occurs during data communication by TCP. In FIG. 14, when a failure occurs in the first relay device 441 at a point 140 p where the transmission of the data with the sequence number 10001 to the second communication device 455 is completed (a point after step S140 to step S142) 140p. It is a failure detection processing flow. In FIG. 14, the ACK packet for the data of sequence number 10001 is not shown.

  The first communication device 454 transmits the data with the sequence number 10001 to the first relay device 441, and then transmits the data with the sequence number 11461 to the first relay device 441 (step S150). Since the first relay device 441 has failed, the received data is transmitted to the second relay device 442 without being processed by the processing unit 48 (step S151). The second relay device 442 that has received the data with the sequence number 11461 detects that the TCP option is not added to the received data and that the ACK number is not subtracted, thereby opposing the first relay device. 441 faults can be detected (step S155). After detecting the failure of the opposing first relay device 441, the second relay device 442 is reset to the first communication device 454 via the second communication device 455 and the first relay device 441 ( RST) packet may be transmitted, and subsequent communication may be forcibly terminated. Further, after detecting the occurrence of a failure in the opposing first relay device 441, the second relay device 442 may transmit the received data to the second communication device 455 by pass-through. When the second relay device 442 transmits the received data to the second communication device 455 by pass-through, the second communication device 455 expects the ACK number of the received data to be added to the main body. If it is different from the value obtained by adding the predetermined value 1000000 to the ACK number, it is determined as a sequence error. Then, the second communication device 455 transmits the expected ACK number (here, ACK number 20001) to the first communication device 454 via the second relay device 442 and the first relay device 441.

  FIG. 15 is a process for explaining a second failure detection mode when a failure occurs during data communication by TCP. In FIG. 15, a failure detection flow when a failure occurs in the second relay device 442 after the transmission of the data with the sequence number 10001 to the second communication device 455 is completed (after step S160 to step S162). It is.

  The first communication device 454 transmits the data with the sequence number 10001 to the first relay device 441, and then transmits the data with the sequence number 11461 to the first relay device 441 (step S170). The first relay device 441 adds a TCP option to the data of the sequence number 11461, and subtracts a predetermined value of 1000000 from the ACK number and transmits the data to the second relay device 442 (step S171). The second relay device 442 passes through the received data (step S172). The second communication device 455 that has received the data from the second relay device 442 has received data with a sequence number different from the expected sequence number (sequence number 1011461). The packet is returned (step S180). The second relay device 442 passes through the ACK packet for the retransmission request because a failure has occurred (step S181). The first relay device 441 detects the failure of the opposing second relay device 442 by detecting that the TCP option is not added to the ACK packet of the retransmission request and that the ACK number is not subtracted. Yes (step S185). After detecting the failure of the opposing second relay device 442, the first relay device 441 is reset to the second communication device 455 via the first communication device 454 and the second relay device 442 ( RST) packet may be transmitted, and subsequent communication may be forcibly terminated. Further, after detecting the failure of the opposing second relay apparatus 442, the first relay apparatus 441 may transmit the received data to the first communication apparatus 454 by pass-through. In FIG. 15, the first relay apparatus 441 detects the occurrence of a failure based on the ACK packet for the data of sequence number 11461 (step S185), but the transmission of the ACK packet for the data of sequence number 10001 is performed more than in step S180. If it is processed first, the occurrence of a failure can be detected based on the ACK packet.

B-4. effect:
As described above, the first and second relay apparatuses 441 and 442 are based on at least one of the TCP option and the ACK number, and one of the relay apparatuses 441 and 442 generates a failure of the other relay apparatus 441 and 442. Can be easily detected. Thereby, for example, when a failure occurs in one of the first and second relay apparatuses 441 and 442, data transmission by pass-through, transmission of an RST packet, etc. without waiting for interruption of data communication due to timeout or the like The processing for advancing data communication can be executed promptly.

  In the above embodiment, the occurrence of a failure in the relay apparatuses 441 and 442 is detected based on the TCP option and the ACK number (for example, FIGS. 10, 12, and 13). In addition to the TCP option, the ACK number is also added to the determination element for detecting the occurrence of the failure, so that the opposite relay device 441, based on the packet to which the TCP option erroneously intruded from another network (connection) is added The possibility that the occurrence of the failure 442 cannot be detected by mistake can be reduced. That is, it is possible to construct the network system 5 that can detect the failure of the relay apparatuses 441 and 442 with higher accuracy.

  Further, in the above-described embodiment, when the detection unit 46 detects the failure of the opposing relay devices 441 and 442 based on the received data, the data received by the relay devices 441 and 442 that have detected the failure includes a SYN packet. In this case, the subsequent data is processed by pass-through (FIG. 8). This makes it possible to establish a connection without interrupting the 3-way handshake process.

  Further, in the above-described embodiment, when the detection unit 46 detects the failure of the opposing relay devices 441 and 442 based on the received data, the data received by the relay devices 441 and 442 that detected the failure is a TCP connection establishment request. When the ACK packet for the SYN packet for performing the process is included, the subsequent data is processed by pass-through (FIG. 9). This makes it possible to establish a connection without interrupting the 3-way handshake process.

  Here, the first relay device 441 and the second relay device 442 of the above-described embodiment correspond to the “first relay device” described in the means for solving the problem. Specifically, in the above-described embodiment, when the first relay device 441 corresponds to the “first relay device” described in the means for solving the problem, the second relay device 442 solves the problem. This corresponds to the “second relay device” described in the means for doing so. Further, in the above embodiment, when the second relay device 442 corresponds to the “first relay device” described in the means for solving the problem, the first relay device 441 solves the problem. This corresponds to the “second relay device” described in the means.

C. Variations:
As mentioned above, although one Example of this invention was described, this invention is not limited to such an Example, A various structure can be taken in the range which does not deviate from the meaning. For example, the following modifications are possible.

C-1. First modification:
In one aspect of the failure detection of the above-described embodiment, a mode in which the failure occurrence of the opposing relay devices 441 and 442 is detected by detecting that the TCP option is not added to the data and that the ACK number is not subtracted is detected. (For example, B-2-4. Third failure detection mode), but is not limited to this. That is, by detecting at least one of the fact that the TCP option is not added to the data and that the ACK number is not subtracted, one of the relay devices 441 and 442 makes the other relay device 441 opposite to the other. The occurrence of a failure 442 may be detected. For example, in FIG. 10, by detecting that the TCP option is not added to the data received by the first relay device 441, the first relay device 441 detects the failure of the opposing second relay device 442. It may be detected. Further, for example, in FIG. 10, the first relay device 441 detects that the ACK number of the data received by the first relay device 441 is not subtracted by a predetermined value, so that the first relay device 441 A failure occurrence may be detected. That is, a relay device having the following modification may be used.

First variant:
Two terminals that transmit data by stream-type communication using sequence numbers, and first and second relays that are arranged in series on the transmission path of the two terminals and relay data between the two terminals A first relay device arranged in a network having a device,
A transfer control unit for transferring received data;
A detection unit that detects that a failure has occurred in the second relay device that faces the second relay device;
The transfer control unit
A transfer unit for transferring the received data;
A processing unit that processes the received data to detect occurrence of a failure in the second relay device, and (i) data to be transferred when data is transferred to the second relay device A processing unit having an optional part for giving a mark to the data to indicate that the data is exchanged between the first and second relay devices,
The 1st relay apparatus containing the said mark as an element for detecting generation | occurrence | production of the failure of the said 2nd relay apparatus which the said detection part opposes.
The first relay device according to the first modified aspect can detect the occurrence of a failure in the opposing second relay device based on the mark. Here, the detection unit can easily detect the occurrence of a failure in the second relay device when a mark is not attached to the data transmitted from the second relay device and received.

Second variant:
Two terminals that transmit data by stream-type communication using sequence numbers, and first and second relays that are arranged in series on the transmission path of the two terminals and relay data between the two terminals A first relay device arranged in a network having a device,
A transfer control unit for transferring received data;
A detection unit that detects that a failure has occurred in the second relay device that faces the second relay device;
The transfer control unit
A transfer unit for transferring the received data;
A processing unit that processes the received data to detect the occurrence of a failure in the second relay device, and (i) transmitted from the terminal that is the transmission source when transmitting data to the opposite terminal An adder that adds a predetermined value to the sequence number assigned to the data to be transmitted, and (ii) the confirmation response number assigned to the data transmitted from the terminal when the data is received from the opposite terminal A processing unit having a subtraction unit for subtracting the predetermined value from
The 1st relay apparatus which contains the said confirmation response number as an element for detecting generation | occurrence | production of the failure of the said 2nd relay apparatus which the said detection part opposes.
The first relay device according to the second modification can detect a failure of the opposing second relay device based on the confirmation response number. Here, when the acknowledgment number is the number obtained by adding a predetermined value to the number that should be received from the data that is transmitted from the second relay device and received, the detection unit may fail the second relay device. Can be easily detected.

C-2. Second modification:
In the above embodiment, the first and second communication devices 454 and 455 perform data communication by TCP, but the protocol used for data communication is not limited to TCP. That is, the present invention can be applied when the first and second communication devices 454 and 455 perform data communication by stream type communication using a sequence number. For example, the present invention can be applied to data communication using RTCP (Real-time Transport Control Protocol).

C-3. Third modification:
In the above embodiment, the first and second relay devices 441 and 442 detect the failure of the opposing relay devices 441 and 442, and the data communication is interrupted by transmitting the RST packet. Thus, data communication may be interrupted. That is, when a control device newly connected to the network system 5 is newly provided and a failure is detected by the relay devices 441 and 442, the relay device that has detected the failure notifies the control device of the failure detection, and information relating to failure detection The control device that has received the data may interrupt the data communication.

5, 5w ... Network system 10w ... First relay device 11w ... Second relay device 20w ... First communication device 21w ... Second communication device 41, 42 ... Input / output interface 44 ... Transfer control unit 46 ... Detection unit 48 ... Processing unit 48a ... Option unit 48b ... Addition unit 48c ... Subtraction unit 49 ... Transfer unit 441 ... First relay device 442 ... Second relay device 454 ... First communication device 455 ... Second communication device

Claims (15)

A first relay device,
A first network interface connected to the first network;
A second network interface connected to the second network;
A transfer unit that relays a packet received via the first network interface in a predetermined transmission band toward the second network via the second network interface;
A bandwidth control unit that controls the predetermined transmission bandwidth based on a communication status between the second relay device and the first relay device;
A packet processing unit configured to process data in a predetermined area in a packet to be relayed when a predetermined transmission band is controlled by the band control unit; The first relay device according to claim 1,
Judgment whether or not the bandwidth control unit controls the transmission bandwidth based on the information given by the second relay device to the packet from the second relay device received via the second network interface A first relay device characterized by that. The first relay device according to claim 1 or 2,
The packet processing unit adds a marker to the predetermined area and processes the data of the predetermined area. The first relay device according to any one of claims 1 to 3, wherein
The bandwidth control unit controls a bandwidth determined according to a retransmission status of a packet from the transfer unit to the second network;
The first relay device, wherein the transfer unit relays a packet in a band controlled by the band control unit or a band according to the TCP. The first relay device according to claim 3,
A first relay device that relays a SYN packet or a SYN / ACK packet by a three-way handshake to establish a connection with the second relay device via the second network interface. . The first relay device according to any one of claims 1 to 5, wherein
The transfer unit relays a packet including a sequence number in the predetermined area according to TCP,
The packet processing unit updates a sequence number included in the predetermined area, wherein the first relay device. A relay device connected to the first base network and the wide area network,
A relay unit that relays a packet from the first base network to the wide area network in a predetermined transmission band;
A bandwidth control unit for controlling the predetermined transmission bandwidth based on the communication status of the wide area network;
A packet processing unit that updates a predetermined area of the header of the packet using information indicating that the packet is relayed in the predetermined transmission band to the external device by the band control unit. A relay device characterized by that. The relay device according to claim 7,
The relay unit relays a packet according to TCP,
The relay apparatus according to claim 1, wherein the bandwidth control unit controls the predetermined transmission bandwidth based on a retransmission state of packets according to TCP in the wide area network. The relay device according to claim 7,
The relay unit relays a packet according to TCP,
The bandwidth control unit stores, in a predetermined area of a header of a packet received from the wide area network, whether or not to control the predetermined transmission band based on a retransmission status of the packet according to TCP in the wide area network. A relay device characterized in that the relay device is determined based on the information. A first bandwidth control unit that is connected via a wide area network to relay packets between communication apparatuses included in different base networks and that controls a transmission band of packets to be relayed based on a communication state in the wide area network. A communication system including the relay device and the second relay device,
The first relay device is
A transfer unit for transferring received packets;
Band control means in the second relay device based on control information provided when relaying the packet in at least one of the first relay device or the second relay device included in the received packet And a determination unit for determining the state of the communication system. The communication system according to claim 10, wherein
The control information includes an identifier indicating that the packet is controlled by the bandwidth control unit of the second relay device,
The said determination part determines that the state of the said band control means is a failure by whether the said identifier is contained in the said received packet, The communication system characterized by the above-mentioned. The communication system according to claim 10, wherein
The first relay device is
The control information includes a sequence number,
Holding means for holding a sequence number included in a packet received from one of the communication devices via a base network;
In response to receipt of the packet to which the control information is added, the determination unit compares the sequence number included in the control information with the held sequence number, thereby controlling the bandwidth of the first relay device The communication system characterized by determining the state of the. A first relay device arranged in a network having two terminals and first and second relay devices arranged on a transmission path of the two terminals for relaying data between the two terminals There,
A bandwidth control unit that controls a bandwidth for transferring received data;
When transferring data to the second relay device, a processing unit that adds to the data a mark for indicating that the data to be transferred is data exchange between the first and second relay devices When,
A transfer unit for transferring the received data;
The second facing each other on the basis of a first mark that is given to data transmitted from the second relay device and indicates that data is exchanged between the first and second relay devices A first relay device comprising: a detection unit that detects occurrence of a failure of the relay device. The first relay device according to claim 13,
The detection unit detects a failure of the second relay device when the first mark is not added to the data transmitted and received from the second relay device. apparatus. The first relay device according to claim 13 or 14,
The processing unit has an update unit that updates at least one of a sequence number or an acknowledgment number assigned to data,
The detection unit includes a first mark added to data transmitted from the second relay device to indicate that the data is exchanged between the first and second relay devices, and a confirmation A first relay device that detects the occurrence of a failure of the opposing second relay device based on at least one of first response numbers that are response numbers.

Images