JP3999785B2 - Communication method - Google Patents

Description

  The present invention relates to a communication method, and more particularly to a communication method for securing a communication path when a NAT function is implemented in IP communication between a client and a server.

  In recent years, due to the increase in computers connected to the Internet, the problem of lack of IP addresses has become prominent. Therefore, a NAT (Network Address Translator) function is used as one method for solving the shortage of IP addresses. The NAT function performs address conversion in order to perform communication between a client computer having a private address in a LAN, which is a closed network, and a host computer having a global address in the Internet.

  FIG. 1 shows a network connection diagram for explaining the NAT function. A private address A and a global address M are assigned to the client 11. The router 13 has a NAT function and stores a conversion table for converting a private address to a global address. Here, the IP packet is transmitted from the client 11 to the host 12 (global address N) in the Internet via the routers 13 and 14. In the header of the IP packet transmitted from the client 11, “N” is written as the destination address 21 and “A” is written as the source address 22, and “200” is set as the destination port number 23 in the TCP header. “100” is written as the number 24. The router 13 refers to the conversion table, converts the private address A of the source address 22 to the global address M, and transmits the packet to the router 14.

  Next, an IP packet is transmitted from the host 12 to the client 11 via the routers 14 and 13. “M” as the destination address 31 and “N” as the source address 32 are written in the header of the IP packet transmitted from the host 12, and “100” is set as the destination port number 33 in the TCP header. “200” is written as the number 34. The router 13 refers to the conversion table, converts the global address M of the destination address 31 to the private address A, and transmits the packet toward the client 11 in the LAN.

  When the NAT function is implemented in the client, the client 11 can connect to the server 12 at an arbitrary timing as described above. However, on the contrary, since the relationship between the global address M, the private address A, and the port number is not specified from the server 12 to the client 11, there is a problem that connection cannot be made at an arbitrary timing.

  In addition, the contents of the conversion table stored in the router are deleted when a certain time elapses after communication is interrupted. That is, the relationship between the destination address, the source address, the destination port number, and the source port number is cleared, and the router 13 does not accept packets with the same address and the same port number.

  Therefore, a method is known in which a TCP session is established from the client 11 to the server 12, and the TCP session is maintained until communication is completed. However, in the server 12, the load for maintaining the TCP session increases, and the number of clients that can be managed by one server is limited to about several thousand. Specifically, because the server 12 stores as many sockets as the number of clients, it consumes a large amount of memory resources and requires CPU resources to search for these sockets.

  The present invention has been made in view of such problems, and an object of the present invention is to provide a communication method capable of connecting a server to a client at an arbitrary timing even in a connection environment using the NAT function. Is to provide.

In order to achieve the above object, the present invention provides a communication method for performing communication between a client and a server connected to a network device having a NAT function. The network device converts the client-side private address and inner port number of the UDP packet received from the client into a network-side global address and outer port number, and transmits the UDP packet to the server. And the network device creates and stores an address conversion table in which the private address and the global address are associated with each other, and generates a port number table in which the inner port number and the outer port number are associated with each other. A second step of storing and the server The said global address and the outer port number of the UDP packet received from the network device, and a third step of storing the set terminal ID and associated map object for each said client, said client, said network device A stealth mode in which a TTL value of the keep-alive packet is a network device closest to the server and does not process an ICMP packet. And a fourth step that is set to a value up to the network device that operates .

The invention according to claim 2 is characterized in that the keep-alive packet according to claim 1 is a UDP echo packet.

According to a third aspect of the present invention, the fixed time interval according to the first or second aspect is less than the time of the no-communication state until the network device determines that the network device is in the no-communication state and closes the UDP port. It is characterized by being set to.

  As described above, according to the present invention, the network device converts the private address and the inner port number into the global address and the outer port number, and transmits the UDP packet to the server. Since the port number is associated with a unique terminal ID for each client, the server can connect to the client at an arbitrary timing even in a connection environment using the NAT function.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the embodiment of the present invention, communication between a client and a server is performed by UDP. By using UDP, the client can be managed without imposing a high load on the server.

  FIG. 2 shows a communication method according to an embodiment of the present invention. A private address A and a global address M are assigned to the client 51. The router 53 has a NAT function and stores an address conversion table for converting a private address to a global address. The client 51 transmits an IP packet to the server 52 (global address N) in the Internet via the routers 53 and 54, and secures a UDP communication path. In the header of the IP packet transmitted from the client 51, “N” as the destination address 61 and “A” as the source address 62 are written, and “200” as the destination port number 63 in the UDP header. “100” is written as the number 64.

  The router 53 refers to the conversion table and converts the private address A of the source address 62 to the global address M. Further, the transmission port number 64 “100” (hereinafter referred to as the inner port number) is converted into the router port number “500” (hereinafter referred to as the outer port number), and then the packet is transmitted to the router 54. At this time, the router 53 generates and stores a port number table in which the inner port number “100” that is the source port number 64 is associated with the outer port number “500” that is the router port number.

  Upon receiving the UDP packet from the client 51, the server 52 stores a map object in which the transmission source IP address “M” and the outer port number “500” are associated with a unique terminal ID for each client. When transmitting data, the server 52 searches for a map of address information using the terminal ID of each client as a key, and generates and transmits a UDP packet.

  A case where an IP packet is transmitted from the server 52 to the client 51 via the routers 54 and 53 will be described. In the IP packet header transmitted from the server 52, “M” is written as the destination address 71 searched for the map from the terminal ID, and the outer port number “500” is written as the destination port number 73 in the UDP header. Further, “N” is written as the source address 72 in the IP packet header, and “200” is written as the source port number 74 in the UDP header. The router 53 refers to the address conversion table and the port number table, converts the global address M of the destination address 71 into the private address A, and converts the outer port number “500” of the destination port number 73 into the inner port number “100”. Then, the packet is transmitted to the client 51 in the LAN.

  Thereafter, the packet can be transmitted from the server 52 to the client 51 at an arbitrary timing using the global address M and the outer port number “500”. Further, as described above, when performing communication using TCP, the server must generate and manage a socket for each session. On the other hand, according to the present embodiment, the server stores the source IP address (global address) and the outer port number using the terminal ID of each client as a key, and searches each time a packet is transmitted. Therefore, the destination can be designated for each packet, and the packet can be transmitted to the client from the combination of the same source IP address and outer port number regardless of the concept of the session. Therefore, memory and CPU resources are not consumed for session management.

  If a firewall that cannot perform UDP communication is connected in the middle of the communication path, communication between the client and server cannot be performed. Therefore, when the UDP packet for network environment investigation is transmitted from the client to the server and the echo packet can be received from the server, the UDP communication according to the present embodiment may be performed.

  As shown in FIG. 2, the UDP communication path generated between the client 51 and the server 52 is regarded as the end of communication if the communication between the client and the server is not performed periodically, and the router May cause the communication path to be closed. The router monitors the communication from the server to the client to maintain the communication path, monitors the communication from the client to the server to maintain the communication path, or monitors the communication in both directions to maintain the communication path. Therefore, a port maintenance packet (hereinafter referred to as a keep alive packet) is periodically transmitted and received between the client and the server. The keep alive packet has the IP header and the UDP header shown in FIG. The routers 54 and 53 recognize that communication is performed between the client 51 and the server 52 by the keep alive packet, and maintain the port number table and maintain the communication path.

  A UDP echo packet may be used instead of the keep alive packet. By regularly exchanging between the client and the server, the communication path can be maintained and disconnection of the communication path can also be detected.

  When all the clients transmit keep-alive packets to the server, the number of packets reaching the server becomes enormous, consuming the bandwidth of the communication path on the server side and consuming CPU resources. Therefore, the client sends keep-alive packets at regular intervals. It is desirable that the time set as the transmission interval is set to be shorter than the time of the non-communication state until it is determined that the router is in the non-communication state and the UDP port is closed. Since the non-communication time varies depending on the router, it is necessary to automatically detect the non-communication time until the port is closed.

FIG. 3 shows a method for determining the transmission interval of keep-alive packets. First, a UDP communication path is secured from the client to the server (see FIG. 2) (S301). Next, the keep-alive packet transmission interval m = r is set (S302), and the keep-alive packet is transmitted / received on this communication path (S303). Initially, r ₀ is executed at intervals as short as a few seconds. When the keep-alive packet transmission / reception process ends (S304), the value of r is increased (S305), and the keep-alive packet transmission / reception process is performed again.

This is repeated, and r is increased until the keep-alive packet processing fails (r _{1 to} r _n ). _If the keep-alive packet processing fails at rn (S304), the keep-alive packet transmission interval m = r _n-1 is set, and the subsequent keep-alive packet transmission / reception processing is performed (S306). This method may be performed at appropriate intervals as needed, and keep-alive packet transmission / reception processing may be performed at the determined transmission interval until the next transmission interval is determined.

  Even if the keep-alive packet transmission interval is adjusted, if all of the keep-alive packets sent from the client to the server arrive at the server, the bandwidth of the communication path on the server side may be consumed and the CPU resources may be consumed. is there. Therefore, the TTL (Time To Live) value of the keep-alive packet transmitted from the client to the server is set to a value that does not reach the server.

  When the TTL value is set to a value that does not reach the server, the router that passes when the TTL value reaches 0 returns an ICMP error message that notifies the transmission source router that the TTL value is insufficient. There is a case. Further, when the transmission source router receives the ICMP error message, the communication path between the client and the server may be closed. Therefore, it is detected whether or not a stealth mode router that does not process the ICMP packet exists in the communication path between the client and the server. By setting the TTL value to the stealth mode router closest to the server, the client can also prevent the occurrence of an ICMP error message and prevent the communication path from being blocked.

  UDP communication does not have a session concept. Therefore, even if the Internet communication path outside the router 53 to which the client 51 is connected is disconnected during communication, the client 51 cannot detect the disconnection. For example, when the IP address of the router 53 is changed after the communication path is reconnected, the client 51 cannot recognize the changed IP address.

  FIG. 4 shows a method for notifying a client of a network change. When the keep-alive packets 81 and 82 are periodically exchanged between the client and the server, the server 52 writes the transmission source address M of the received keep-alive packet 81 in the keep-alive packet 82. The client 51 refers to the IP address information in the keep-alive packet sent from the server 52, and can recognize that the network has been changed if the source address has changed (M → P). . Note that a UDP echo packet may be used instead of the keep alive packet, as in the case of detecting the disconnection of the communication path.

  In IP communication between a client and a server using UDP, data may be lost due to occurrence of packet loss. Accordingly, as in the case of TCP communication, it is possible to confirm that the message has been delivered reliably by returning a confirmation signal (hereinafter referred to as ACK). If an ACK is not returned for a certain period of time despite the transmission of a UDP packet, a retransmission process can be performed. Since a large amount of retransmission processing consumes a large amount of network bandwidth, if retransmission is performed several times for the same UDP packet, the second retransmission from the first time and the third retransmission from the second time Increase the time interval. In addition, an upper limit of the number of retransmissions is set, and when the number of retransmissions exceeds the upper limit, the process ends as a packet missing.

  The present invention can be applied to network devices such as clients, servers, routers, and gateways that can realize IP communication between a client and a server even in a connection environment using the NAT function.

It is a network connection diagram for demonstrating a NAT function. It is a network connection figure which shows the communication method concerning one Embodiment of this invention. It is a flowchart which shows the method of determining the transmission interval of a keep alive packet. It is a network connection diagram which shows the method of performing a network change notification to a client.

Explanation of symbols

11, 51 Client 12, 52 Server 13, 14, 53, 54 Router

Claims (3)

In a communication method for performing communication between a client and a server connected to a network device having a NAT function,
The network device converts the client-side private address and inner port number of the UDP packet received from the client into a network-side global address and outer port number, and transmits the UDP packet to the server. Steps,
The network device creates and stores an address conversion table that associates the private address with the global address, and generates and stores a port number table that associates the inner port number with the outer port number. The second step;
A third step in which the server stores the global address and the outer port number of the UDP packet received from the network device in a map object associated with a terminal ID set for each client ;
The client sends a keep-alive packet to the server via the network device at a predetermined time interval, and the TTL value of the keep-alive packet is the network device closest to the server. And a fourth step set to a value up to a network device operating in a stealth mode that does not process an ICMP packet . The communication method according to claim 1 , wherein the keep-alive packet is a UDP echo packet. Said predetermined time interval, the network device, determines that the non-communication state, according to claim 1 or 2, characterized in that it is set to less than the time of non-communication state before closing the UDP port Communication method.

Images