JP4839428B2 - Multi-home connection method and apparatus - Google Patents

Description

  The present invention relates to a service (hereinafter referred to as multihomed connection) that connects to a plurality of Internet service providers (hereinafter referred to as ISPs), and more particularly, to control a packet transfer path from the Internet to a user network and to perform multihomed connection. It is related with the multihome connection method and apparatus which can implement | achieve easily.

  In the Internet connection service, the multi-home connection is a mechanism for connecting to a plurality of ISPs and making the connection with the Internet redundant. When the user network is connected to only one ISP, as shown in FIG. 11, when one ISP-A fails in the user network 4 connected to the external client 3 connected via the Internet 2, the user network The connection between 4 and the Internet 2 is interrupted. In order to prevent this, the user network 4 is connected to another ISP-B or the like so that even when one ISP fails, the connection to the Internet is maintained via the other ISP. It becomes possible to do. In particular, a network in which an Internet connection is directly connected to a business, such as an EC (Electronic Commerce) site, requires a multi-home connection.

In addition, as disclosed in Patent Documents 1 and 2, for a multi-home connected user network, traffic from the Internet is appropriately distributed over a plurality of links, and communication quality parameters such as delay and packet loss. Accordingly, there has been proposed a packet transfer method in a multihome connection in which a route passing through any one of a plurality of links can be appropriately selected.
JP 2004-274702 A JP 2004-274703 A

  In the above conventional technique, in order to realize multihome connection, a mechanism for dynamically avoiding the use of the ISP in which a failure has occurred is necessary. Even if connected to a plurality of ISPs, if there is no mechanism for detecting a failure of the ISP and switching to use the other ISP, the ISP that has failed will continue to be used. Therefore, it is necessary to have a mechanism for detecting a failure of the ISP and using the other ISP as a trigger.

  In addition, in multi-homed connection, not only redundancy of connection but also a mechanism for distributing transmission / reception of traffic to a plurality of ISPs and improving line utilization efficiency is required. When the traffic distribution is not controlled, there is a possibility that the traffic is concentrated on one ISP and the line is congested even though the other ISP has a surplus bandwidth. For this reason, it is preferable to improve the utilization efficiency of the line by introducing a mechanism for appropriately using each line even during normal times.

  In general, a mechanism called BGP (Border Gateway Protocol) is widely used to realize multi-home connection, but some problems have been pointed out. First, the ISP cannot be selected based on the communication quality. In the case of BGP, a situation in which communication with the ISP is not possible is used as a failure criterion. Therefore, when a slight packet loss occurs or when the delay becomes large, it is not regarded as a failure and the ISP continues to be used. As a result, actual communication is exchanged over a line with poor quality. Moreover, since control according to the traffic volume cannot be performed, in order to perform appropriate traffic distribution, it is necessary for the administrator to always monitor and manually control the traffic volume. In addition to these technical issues, BGP has a problem of high operational costs, and it has been difficult to introduce this BGP to a small network such as a small business.

  In addition, multi-NAT (Network Address Translation), which is a conventional multi-network address translation function, has a problem that sessions originating from an external network become inconsistent. Only those sessions have been used for redundancy.

  The present invention has been made in view of the above-mentioned problems of the prior art, solves problems such as BGP with a simple structure, enables multi-home connection easily, and makes efficient Internet connection high. An object of the present invention is to provide a multihome connection method and apparatus capable of maintaining communication quality.

  The multi-home connection method and apparatus according to the present invention is a multi-home connection apparatus that dynamically avoids a failure by using a redundant system for connecting users and ISPs and a redundant system for connecting ISPs and the Internet.

In the present invention, a center tunnel router and an interface are installed in each of a plurality of Internet service providers, a user tunnel router is installed on the user side of the Internet service provider, and the user tunnel router is the center tunnel router. A logical circuit that connects to an Internet connection service provided by each Internet service provider installed in the network and passes physically different interfaces and paths between the user and each tunnel router of each Internet service provider. And each Internet service provider is interconnected by a network, performs access line control based on steady communication quality measurement of each logical line by the user, and which logical line Whether to use for communication is always selected dynamically, and each tunnel router compares the result of quality measurement of each logical line, controls the route to use the logical line judged to be optimal, and A network address translation router and a domain name system server are provided at each site, and the user 's network address translation router corresponds to each of the plurality of Internet service providers, and each of the Internet service providers virtually corresponds to each of the virtual globals. receiving the address assignment, the private network of the user is constructed by a private address of the user, the private network of the user, in association with the private address and the virtual global address in a many-to-one, said different virtual Access to global address, and can be transferred to one of the private server of the user, the registered plurality of virtual global address to the private server of the domain name, when in one of the Internet service provider fails The domain name system server does not respond to the domain name address inquiry by returning the virtual global address assigned by the Internet service provider in which the failure has occurred , thereby allowing access via the Internet service provider having a failure. avoiding, the network address translation router, when determining the next-hop to a destination port which preferentially against the next hop registered in the normal routing table, set by the user Performs policy routing to select the next hop based on Rishi, the network address translation router, the default provided with a routing table in which a predetermined destination is set as the destination, provided the flow table and the policy routing table, the network address Translation router, when receiving a packet from outside, the source address of the packet from the external destination address, source port number, information about the destination port number, flow entries with external information of the interface that received the registers in the flow table, during a predetermined time set arbitrarily, if it does not receive the packet from other external matching the registered condition as entries in the flow table, determine that the communication has disappeared as And deletes the entry from the flow table, when receiving a packet from other external matching the entry in the timer time, the timer of the entry is reset, if it detects a down interface, immediately The related entries are deleted, and the policy routing table is a table for determining the interface of the next hop to be transferred on condition of the destination address of the packet to be transferred , the source port number, and the information of the destination port number. The flow entry by the external packet registered in the flow table is dynamically added to the policy routing table and set, and the policy routing table is always searched before the routing table, War Address translation router, if the entry of the packet to the forwarding the policy routing table matches is a multi-homed connection to select the next hop registered in the policy routing table as the destination of the packet to the transfer .

The present invention comprises a plurality of Internet service providers and their users, each Internet service provider side is provided with a center tunnel route and an interface , and each user side is provided with a user tunnel router, Each of the user tunnel routers is connected to an Internet connection service provided by each Internet service provider in which the center tunnel router is installed, and each of the user and each Internet service provider is connected using a physical line. A logical circuit is established between each tunnel router, and each Internet service provider is interconnected by a network, and a physical connection is established between the user and the tunnel router of one Internet service provider. Establishes each multiple logical lines through different said interface and route, the communication quality measuring means for performing communication quality measurement of said logic circuits, each equipped wherein the tunnel router and tunnel router for user for each center, the communication The quality measuring means constantly measures the quality of each logical line established between the user and each tunnel router of each Internet service provider, and constantly dynamically uses a more efficient logical line for communication. A multi-home connection apparatus that selects and controls a path to use each of the logical lines. The user has a private network, and the user base includes a network address translation router and a domain name system server. The user network address translation router is connected to the plurality of Internet service providers, and each Internet Upon receiving an address assignment from a service provider, the user's private network is constructed with the user's private address, and a plurality of virtual global addresses are virtually assigned to the user's private server installed in the user's private network. The virtual global address and the private address are set in a one-to-one correspondence, and access to the different virtual global addresses is assigned to one private service. The domain name system server registers a plurality of the virtual global addresses for the domain name of the private server, and when a failure occurs in one Internet service provider, the domain name The system server avoids access via the faulty Internet service provider by not returning the virtual global address assigned by the faulty Internet service provider in response to the domain name address inquiry. The network address translation router includes a routing table in which a predetermined destination is set as a default destination, and includes a flow table and a policy routing table. When the address translation router receives a packet from the outside, the information about the source address, the destination address, the source port number, and the destination port number of the packet from the outside is received as a flow entry together with the received external interface information. If it is registered in the flow table and a packet from the outside that matches the condition registered as an entry in the flow table is not received for an arbitrarily set fixed time, it is determined that communication is lost, If the entry is deleted from the flow table and another external packet matching the entry is received within the timer time, the timer of that entry is reset, and if an interface down is detected, the relevant entry is immediately Delete the policy routing The table is a table for determining the interface of the next hop to be transferred on the condition of the destination address of the packet to be transferred, the source port number, and the information of the destination port number, and from the outside registered in the flow table The flow entry by the packet is dynamically added to the policy routing table and set, the policy routing table is always searched before the routing table, and the network address translation router When the entry of the policy routing table matches, the multi-home connection apparatus selects the next hop registered in the policy routing table as a destination of the packet to be transferred .

In addition, the present invention has a private network connected to a plurality of Internet service providers, and each Internet service provider side is provided with a center tunnel router and an interface corresponding to the connection destination of each of the users having the private network. A network address translation router and a domain name system server are provided at the base, and the user network address translation router is connected to a plurality of the Internet service providers, receives an address assignment from each Internet service provider, and receives the user's private network Is constructed by the user's private address, and the user's private server installed in the user's private network includes a plurality of Virtual global address is virtually set, the correspondence of the virtual global address and the private address in a many-to-one, access to different ones of the virtual global address, and can be transferred to one the private server, the domain name system The server registers a plurality of the virtual global addresses for the domain name of the private server, and when a failure occurs in one Internet service provider, the domain name system server issues an address query for the domain name. against it, that does not respond to the virtual global address assigned by the Internet service provider to which the failure occurs, is intended to avoid the access via the Internet service provider with disabilities, Serial network address translation router is provided with a routing table that predetermined destination is set as the default destination, a flow table and policy routing table, the network address translation router, when receiving a packet from outside, the Information related to the source address, destination address, source port number, and destination port number of the packet from the outside is registered in the flow table as a flow entry together with the received external interface information, and the information is set for an arbitrarily set fixed time. If no other external packet that matches the condition registered as an entry in the flow table is received, it is determined that communication is lost, the entry is deleted from the flow table, and the timer is expired. Ent When receiving a packet from other external matching the re timer of the entry is reset, if it detects a down interface, delete the immediately associated entry, the policy routing table, the packet to be transferred A table for determining the interface of the next hop to be transferred on condition of destination address, source port number, and destination port number information, and the flow entry by the external packet registered in the flow table. , and set to dynamically add to the policy routing table, the policy routing table is always searched before the said routing table, the network address translation router, the policy routing tape packet to the transfer If Le entry matches a multihomed device for selecting the next hop registered in the policy routing table as the destination of a packet to be the transfer.

  The plurality of Internet service providers have a configuration in which logical line connections are symmetric with respect to the user.

  In the system of the present invention, an optimum line can be selected on the basis of bidirectional quality measurement between a user and an ISP. Then, a policy routing technique is used for controlling so that a response packet is sent back to the same line as the access when there is an access from an external network. By using this technology, the multi-NAT function is used for bidirectional redundancy.

  According to the multihome connection method and apparatus of the present invention, high-speed line failure detection and line switching are possible, and the line failure is bypassed at high speed. You can continue to use the Internet exactly as you did.

  It is possible to automatically perform line switching based on communication quality. That is, when packet loss or delay occurs, it is determined as a failure, and control is performed so that communication is performed using another line. Thereby, a user's traffic can always maintain high quality.

  In addition, the line can be switched according to the traffic volume, and when the traffic volume flowing over the line exceeds the set threshold, the traffic can be switched to another line in order to prevent congestion in advance. . Appropriate traffic distribution happens automatically, so network administrators do not need to perform maintenance.

  Accurate communication quality measurement is possible, and quality information is obtained by measurement, and measurement is performed in both directions from the user and ISP, so it is possible to measure the quality in each direction and grasp the more accurate quality. It becomes possible.

  Switching can be performed while maintaining a session, and an existing session can be maintained even when a failure occurs. For this reason, the user side is not affected by the failure. Furthermore, multihoming using a PA (Provider Aggregatable) address with a low running cost is possible.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 shows a first embodiment of the present invention. In a system using the multihome connection method and apparatus of this embodiment, as a failure pattern, a failure in connection between a user and an ISP, and a connection between the ISP and the Internet. Assumes failure. Therefore, the present invention dynamically avoids a failure by (A) a redundant system for connection between the user and the ISP and (B) a redundant system for connection between the ISP and the Internet. Both of these can be used independently, but by combining the two, optimal redundancy of the Internet connection is realized.

(A) Redundant System for Connection between User and ISP Here, a method for making the connection line between the user 12 and the ISP redundant will be described. As shown in FIG. 1, the system according to this embodiment includes a plurality of tunnel routers. A center tunnel router 10 is installed in ISP-A, and a user tunnel router 14 is installed on the user 12 side. The tunnel router 14 of the user 12 is logically connected to the ISP-A tunnel router 10 and always communicates with the Internet 20 via the ISP-A tunnel router.

  The tunnel router 14 of the user 12 is redundantly connected to the ISP-A tunnel router 10 using a plurality of logical lines based on IP (Internet Protocol) tunnel technology. The IP tunnel technology is a technology for establishing a virtual logical line between routers via an IP network. GRE and IP-In-IP are standardized.

  First, the user 12 makes a contract with a general Internet connection service such as ADSL or FLETS provided by one ISP-A in which the tunnel router 10 is installed. Using this IP connection line 11A, a logical line 13A by tunnel is established between the user 12 and the ISP tunnel router.

  The user 12 contracts with another ISP-B for an Internet connection service and uses this IP connection line 11B to establish another logical line 13B between the user 12 and a tunnel router (not shown) of the ISP-B. Establish. ISP-A and ISP-B are interconnected by an IP network 15. As a result, a plurality of logical lines 13A and 13B are established between the user 12 and the tunnel router 10 of the ISP-A, which physically pass through different paths. Thereby, even when a failure occurs in the IP connection line 11B which is one physical line, the connectivity between the user 12 and the ISP-A is maintained.

  The routers 10 and 14 regularly monitor the logical lines 13A and 13B by alive confirmation using a keep alive packet, and when a failure is detected in the logical lines 13A and 13B, the traffic is not used. To detour. By adjusting the life and death confirmation interval, detours on the order of several seconds are possible.

  In the case of this configuration, the network of the user 12 is a network belonging to ISP-A. ISP-B functions as an access line provider for connecting user 12 and ISP-A.

  Next, access line control based on logical line quality measurement will be described. First, the user 12 and the ISP-A tunnel routers 14 and 10 periodically measure the quality of the established logical lines 13A and 13B, and based on the results, which logical lines 13A and 13B are used for communication. Dynamically select what to do. Each of the tunnel routers 10 and 14 internally compares the quality measurement results of the logical lines 13A and 13B, and controls the path so as to use the logical line determined to be optimal according to the operator's policy. Any existing technique may be used for quality measurement of the logical lines 13A and 13B. It is also possible to use existing measurement techniques such as Ping.

(B) Redundant System for Connection between ISP and Internet Next, a method for making the private network 26 redundant when a failure occurs in a line between the ISP and the Internet 20 will be described. As shown in FIG. 3, the system according to this embodiment includes a network address translation (NAT) router 22 which is a network address translation router installed at the base of the user 12, and a DNS (Domain
Name System) server 24. The NAT router 22 of the user 12 is connected to a plurality of ISP-A, ISP-B, etc., and receives address assignment from each ISP. The private network 26 of the user 12 is constructed with the private address of the user 12.

  In the present invention, the private network 26 of the user 12 is constructed using so-called multi-NAT, thereby realizing redundancy when a failure occurs in the connection between the ISP and the Internet 20. Multi-NAT is a function implemented in the NAT router 22 of the user 12 and is a mechanism for virtually setting a plurality of global addresses in the private server 28 installed in the private network 26.

  When a general NAT accesses a virtual global address held in the router, the NAT transfers the access to the user's private address associated with the global address on a one-to-one basis. When forwarding, the destination address of the packet is converted into a private address. This mechanism makes it possible for a computer on the private network and an external computer to communicate seamlessly.

On the other hand, in multi-NAT, a virtual global address and a private address that are virtually allocated can be associated one-to-one to transfer access to a different virtual global address to one private server 28 of the user 12. It becomes possible. From the external network, one private server 28 appears as two independent virtual servers 28A and 28B. Using this feature, for one of the private server 28, a virtual global address assigned by the ISP-A, the virtual global address assigned by the ISP-B, assigns each virtually. In this way, even if a failure occurs in one ISP-A, it is possible to continue communication with the private server 28 via the other virtual global address.

In the case of this configuration, since there are a plurality of virtual global addresses for one server 28, the DNS server 24 registers a plurality of virtual global addresses for the domain name of this server 28. When a failure occurs in one ISP, the DNS server 24 avoids access via the failed ISP by not returning the virtual global address assigned by the ISP in response to the domain name address inquiry. .

  Generally, when multi-NAT is used, TCP (Transmission Control Protocol) session establishment from the outside becomes a problem. When communication from the inside to the outside is performed in the NAT, the source address of the packet is converted from the private address to the global address and transferred to the destination. In the case of multi-NAT, since a plurality of global addresses are associated with one private address, it is necessary to select which global address is used as a source address. In general routers, a global address to be converted is selected according to an interface to which a packet is transferred. For example, when transferring to ISP-A, it is converted to a global address associated with the interface connected to ISP-A, and when transferring to ISP-B, it is associated with the interface connected to ISP-B. To the specified global address.

  Here, based on FIG. 4, consider a case where an external client 30 on the Internet accesses the virtual server 28B by TCP. Further, it is assumed that the NAT router 22 is set by default to forward traffic addressed to the external client 30 to the ISP-A. In this case, the transmission source address of the response packet from the private server 28 to the external client 30 is converted to that of the virtual server 28A. Since the server of the external client 30 is waiting for a response from the virtual server 28B, the response from the virtual server 28A is not processed. As a result, communication between the private server 28 and the external client 30 is not established.

  Therefore, the system according to the embodiment of the present invention performs policy routing based on the session log in order to solve this problem. The policy refers to route selection conditions other than the destination IP address. For example, the source IP address, the destination port number, or a combination thereof can be cited as a policy.

  First, routing on the Internet is based on the destination base. Using the destination IP address information included in the IP header of the packet as a key, the destination entry registered in its own routing table is searched, and the next hop corresponding to it is selected. On the other hand, in policy routing, when determining the next hop to a certain destination, the next hop is selected based on the policy set by the user, not the next hop registered in the normal routing table.

In the system according to the embodiment of the present invention, the policy is dynamically set based on the session log. Upon receiving access from the external client 30, a policy for dynamically changing the next hop addressed to the external client 30 to the interface that has received access (hereinafter referred to as IF) is dynamically added. As in the above example, when the external client 30 accesses the virtual server 28 B, the NAT router 22 changes the next hop addressed to the external client 30 from ISP-A of the initial setting to ISP-B. Dynamically add policies to As a result, as shown in FIG. 5, the source address of the response packet is converted into the address of the virtual server 28B, and the TCP session is correctly established.

  The policy routing of this embodiment will be described in detail below. As shown in FIG. 6, in the NAT router 22, an interface connected to ISP-A is IF-A, and an interface connected to ISP-B is IF-B. As IP addresses, Addr-A2 is set for IF-A and Addr-B2 is set for IF-B. The address of the ISP-A router is Addr-A1, and the address of the ISP-B router is Addr-B1.

  In the private server 28 inside the user 12, Addr-X2 is set as the IP address. Further, the interface of the NAT router 22 connected to the private server 28 is IF-X, and Addr-X1 is set as an address.

  In the routing table 34 of the NAT router 22, Addr-A1 is set as a default destination. The routing table 34 is a database table for determining a packet transfer destination. Search for a matching destination using the destination address of the packet as a key. This setting represents that all the packets transferred by the NAT router 22 are normally transmitted toward Addr-A1, that is, ISP-A. In the NAT table 32A corresponding to IF-A, the association between the internal address Addr-X2 and the external address Addr-A2 is registered.

The packet passing through IF-A is converted in address according to the NAT table 32A. Further, the association between the internal address Addr-X2 and the external address Addr-B2 is registered in the NAT table 32B corresponding to IF-B. A packet passing through IF-B is converted in address according to the NAT table 32B.

  Next, a basic NAT address translation process will be described. First, when transferring a packet from the outside to the inside, as shown in FIGS. 7 and 8, the processing process of the NAT router 22 is as follows.

  A packet is received from the outside (S1). Then, the NAT table 32A of the received interface is referred to (S2). If the destination address of the packet matches any of the external addresses registered in the NAT table 32A, the NAT router 22 converts the destination address of the packet into the corresponding internal address Addr-X2 (S3).

Next, the routing table 34 is referred to. Based on the information of the changed destination address Addr-X2, it searches the IF-X of the next hop (S4). Then, the packet is transferred to IF-X of the next hop , and the packet with the destination address converted is transmitted from IF-X to private server 28 (S5).

Further, when the packet is transferred from the inside to the outside when the packet is transferred from the inside to the outside (S6), the NAT router 22 first references the routing table 34 (S7). Then, based on the information of the destination address, to search for IF-A of the next hop (S7). According to the next hop, which interface's NAT table 32 is to be referred to is determined (S8). When the packet source address Addr-X2 matches any of the internal addresses registered in the NAT table 32, the NAT router 22 sets the packet source address Addr-X2 to the external address Addr-A2 corresponding thereto. (S9). Then, transfer the packet to the IF-A of the next hop, and transmits the packet to translates the source address from the IF-A is an interface of the next hop.

  Furthermore, in the present invention, the following table is added to the NAT router 22 as shown in FIG. 9 for dynamic policy routing.

First, a flow table 36 is added. The flow table 36 is a table for grasping packets received from the outside. When the NAT router 22 receives a packet from the outside, the NAT router 22 sends information on the source address, destination address, source port number, and destination port number of the packet from the outside as a flow entry to the flow table 36 together with information on the reception interface. sign up. Each entry in the flow table 36 is managed by a timer. If no other external packet that matches the condition registered as an entry in the flow table 36 is received for an arbitrarily set time, it is determined that communication is lost and the entry is deleted from the flow table 36. . If another external packet that matches the entry is received within the time of the timer, the timer for that entry is reset. If an interface down is detected, the related entry is immediately deleted.

Second, a policy routing table 38 is added. The policy routing table 38 is a table for determining the next hop interface to be transferred on condition of the destination address, source port number, and destination port number information of the packet to be transferred.

If the packet to be transferred matches the entry in the policy routing table 38, the next hop interface registered in the policy routing table 38 is selected. The entry of the policy routing table 38 is dynamically added by the flow table 36 described above . When an entry is added to the flow table 36, a policy routing entry is created by replacing the information of the transmission source and the destination based on the information. When the entry of the flow table 36 is deleted, the entry of the related policy routing table 38 is similarly deleted.

  The policy routing table 38 has priority over the normal routing table 34. The policy routing table 38 is always searched before the routing table 34. If the packet matches the conditions of the policy routing table 38, the routing table 34 is not searched.

  When the policy routing described above is performed and the next hop to a certain destination is determined, the next hop is selected based on the policy set by the user, not the next hop registered in the normal routing table, and dynamically When the multi-NAT is used, the connection between the ISP and the Internet 20 can be made redundant.

  In the multihome connection method and apparatus according to the embodiment of the present invention, the center tunnel router 10 is installed in the ISP, the user tunnel router 14 is installed on the user 12 side, and the tunnel router 14 of the user 12 is ISP-A. The tunnel router 10 is logically connected and always communicates with the Internet 20 via the ISP-A tunnel router 10. Furthermore, the network of the user 12 is constructed using multi-NAT, and redundancy is performed when a failure occurs in the connection between the ISP and the Internet 20.

  And in this embodiment, as shown in FIG. 10, the system demonstrated by the above-mentioned (A) and (B) is cooperated, and total multihome service is implement | achieved. That is, the router installed at the user 12 side base is a multi-home router 40 having both functions of the tunnel router 14 and the NAT router 22. In ISP-A and ISP-B, a multihome router 42 having the function of the tunnel router 16 for multihome is installed. Then, between the user 12 and ISP-A and ISP-B, logical lines 13A1 and 13B1 and logical lines 13A2 and 13B2 connected to the Internet 20 via the other ISP are established. This makes it possible for ISP-A and ISP-B to have the same functions as each other, and even when a failure occurs in one of the physical lines, between ISP-A and ISP-B. In addition, at least one logical line can be maintained. This connection model is based on a model standardized as RF2260 and RFC3178 in IETF.

  Note that the multihome connection method and apparatus of the present invention are not limited to the above-described embodiments, and the number of ISPs and NAT routers, installation positions, and the like can be selected as appropriate.

It is a conceptual diagram which shows the connection of a user and ISP in the multihome connection of one Embodiment of this invention. It is a conceptual diagram which shows the connection of a user and ISP in the multihome connection of one Embodiment of this invention. It is a conceptual diagram which shows the connection of a user, ISP, and the internet in the multihome connection of one Embodiment of this invention. It is a conceptual diagram which shows the connection of a user, ISP, and the internet in the multihome connection of one Embodiment of this invention. It is a conceptual diagram which shows the connection of a user, ISP, and the internet in the multihome connection of one Embodiment of this invention. It is a conceptual diagram which shows the system by the side of the user of the multihome connection apparatus of one Embodiment of this invention. It is a conceptual diagram which shows the function of the NAT router of the system by the side of the user of the multihome connection apparatus of one Embodiment of this invention. It is a flowchart which shows the general process of the NAT router of the system by the side of the user of the multihome connection apparatus of one Embodiment of this invention. It is a flowchart which shows the process of the NAT router of the system by the side of the user of the multihome connection apparatus of one Embodiment of this invention. It is a conceptual diagram which shows the multihome connection apparatus of one Embodiment of this invention. It is a conceptual diagram which shows the connection state of the conventional internet and ISP.

10, 14 Tunnel router 11A, 11B IP connection line 12 User 13A, 13B Logical line 15 IP network 20 Internet 24 DNS server 26 Private network 28 Private server 30 External client

Claims (4)

Set up a tunnel router and interface for each center in multiple Internet service providers,
A user tunnel router is installed on the user side of the Internet service provider,
The user tunnel router is connected to an Internet connection service provided by each Internet service provider in which the center tunnel router is installed,
A logical circuit is established between the user and each tunnel router of each Internet service provider through the physically different interfaces and paths,
Each Internet service provider is interconnected by a network, and the user performs access line control based on steady communication quality measurement of each logical line, and dynamically selects which logical line is used for communication. And
Each tunnel router compares the results of the quality measurement of each logical channel, and control passes to utilize the logical channel it is determined that the optimum,
A network address translation router and a domain name system server are provided at the base of the user, and the network address translation router of the user corresponds to each of the plurality of Internet service providers and virtually corresponds to each of the Internet service providers . Assigned virtual global address,
Private network of the user is constructed by a private address of the user, the private network of the user, in association with the private address and the virtual global address in a many-to-one, access to different ones of the virtual global address, and it can be transferred to one of the private server of the user,
When a plurality of virtual global addresses are registered for the domain name of the private server and one of the Internet service providers fails, the domain name system server responds to the domain name address inquiry by by not respond to the virtual global address assigned by the Internet service provider of a failure, to avoid access via an Internet service provider with disabilities,
When determining the next hop to a certain destination, the network address translation router selects the next hop based on the policy set by the user in preference to the next hop registered in the normal routing table. Policy routing,
In addition to providing a routing table in which a predetermined destination is set as a default destination in the network address translation router , a flow table and a policy routing table are provided,
The network address translation router, when receiving a packet from outside, the source address of the packet from the external destination address, source port number, the flow of information about the destination port number, together with the external information of the interface that received If it is registered in the flow table as an entry and no external packet that matches the conditions registered as an entry in the flow table is received for an arbitrarily set time, it is determined that communication is lost. If the entry is deleted from the flow table and another external packet matching the entry is received within the timer period, the timer of the entry is reset, and if an interface down is detected, Delete the entry
The policy routing table is a table for determining the interface of the next hop to be transferred on the condition of the destination address, source port number, and destination port number information of the packet to be transferred , and is registered in the flow table. The flow entry by the external packet is dynamically added and set in the policy routing table, and the policy routing table is always searched before the routing table,
The network address translation router, if the entry of the packet to the forwarding the policy routing table matches is a feature to select the next hop registered in the policy routing table as the destination of the packet to the transfer Multi-home connection method. Consists of multiple Internet service providers and their users,
Each Internet service provider side is provided with a center tunnel router and interface , and the user side is provided with a user tunnel router,
Each of the user tunnel routers is connected to an Internet connection service provided by each Internet service provider in which the center tunnel router is installed, and each of the user and each Internet service provider is connected using a physical line. Establish logical lines between each tunnel router,
Each Internet service provider is interconnected by a network, and establishes each of a plurality of logical circuits through the physically different interfaces and paths between the user and the tunnel router of one Internet service provider,
Each of the center tunnel router and the user tunnel router includes communication quality measurement means for measuring the communication quality of the logical line, and the communication quality measurement means is provided for each tunnel router of the user and each Internet service provider. Measure the quality regularly for each logical line established between them, and always select dynamically to use a more efficient logical line for communication, and control the path to use each logical line Is,
The user has a private network, and the user base includes a network address translation router and a domain name system server,
The network address translation router of the user is connected to a plurality of the Internet service providers, receives an address assignment from each Internet service provider,
Private network of the user is constructed by a private address of the user, wherein the private server of the user who placed the private network of the user, a plurality of virtual global address is virtually set, the said virtual global address associating private address in a many-to-one, access to different ones of the virtual global address, and can be transferred to one the private server,
The domain name system server registers a plurality of the virtual global addresses for the domain name of the private server, and when a failure occurs in one Internet service provider, the domain name system server relative address query that does not respond to the virtual global address assigned by the Internet service provider to which the failure occurs, is intended to avoid the access via the Internet service provider with disabilities,
The network address translation router includes a routing table in which a predetermined destination is set as a default destination , and includes a flow table and a policy routing table.
The network address translation router, when receiving a packet from outside, the source address of the packet from the external destination address, source port number, the flow of information about the destination port number, together with the external information of the interface that received If it is registered in the flow table as an entry and no external packet that matches the conditions registered as an entry in the flow table is received for an arbitrarily set time, it is determined that communication is lost. If the entry is deleted from the flow table and another external packet matching the entry is received within the timer period, the timer of the entry is reset, and if an interface down is detected, Delete the entry
The policy routing table is a table for determining the interface of the next hop to be transferred on the condition of the destination address, source port number, and destination port number information of the packet to be transferred , and is registered in the flow table. The flow entry by the external packet is dynamically added and set in the policy routing table, and the policy routing table is always searched before the routing table,
Said network address translation router, if the entry of the packet to the forwarding the policy routing table matches is to select the next hop registered in the policy routing table as the destination of the packet to the transfer A multi-home connection device characterized by. Have a private network connected to multiple Internet service providers,
Each Internet service provider side has a tunnel router for the center and an interface corresponding to the connection destination ,
At the base of the user having the private network, a network address translation router and a domain name system server are provided,
The network address translation router of the user is connected to a plurality of the Internet service providers, receives an address assignment from each Internet service provider,
Private network of the user is constructed by a private address of the user, wherein the private server of the user who placed the private network of the user, a plurality of virtual global address is virtually set, the said virtual global address associating private address in a many-to-one, access to different ones of the virtual global address, and can be transferred to one the private server,
The domain name system server registers a plurality of the virtual global addresses for the domain name of the private server, and when a failure occurs in one Internet service provider, the domain name system server relative address query that does not respond to the virtual global address assigned by the Internet service provider to which the failure occurs, is intended to avoid the access via the Internet service provider with disabilities,
The network address translation router includes a routing table in which a predetermined destination is set as a default destination , and includes a flow table and a policy routing table.
The network address translation router, when receiving a packet from outside, the source address of the packet from the external destination address, source port number, the flow of information about the destination port number, together with the external information of the interface that received If it is registered in the flow table as an entry and no external packet that matches the conditions registered as an entry in the flow table is received for an arbitrarily set time, it is determined that communication is lost. If the entry is deleted from the flow table and another external packet matching the entry is received within the timer period, the timer of the entry is reset, and if an interface down is detected, Delete the entry
The policy routing table is a table for determining the interface of the next hop to be transferred on the condition of the destination address, source port number, and destination port number information of the packet to be transferred , and is registered in the flow table. The flow entry by the external packet is dynamically added and set in the policy routing table, and the policy routing table is always searched before the routing table,
Said network address translation router, if the entry of the packet to the forwarding the policy routing table matches is to select the next hop registered in the policy routing table as the destination of the packet to the transfer A multi-home connection device characterized by.   4. The multi-home connection apparatus according to claim 2, wherein the plurality of Internet service providers have a configuration in which logical line connections are symmetric with respect to the user.

Images