JP4579746B2 - Connection management device, connection management device control method, and control program - Google Patents

Description

  The present invention relates to a connection management device, a connection management device control method, and a control program, and in particular, a connection management device that exchanges route information in accordance with BGP (Border Gateway protocol: RFC1966), which is a connection communication type route information exchange protocol, The present invention relates to a control method and a control program for a connection management device.

  Conventionally, when constructing a BGP session between two routers when constructing a network system, there is a layer 2 switch (L2 switch), a layer 3 switch (L3 switch) or another router between the two routers. One router cannot directly detect an abnormality in a connection path between the other router and a layer 2 switch (L2 switch), a layer 3 switch (L3 switch), or another router.

  Therefore, in the conventional network system, in order to cope with such a connection path abnormality that cannot be directly detected, a hold timer for disconnecting a BGP session when a period during which transmission could not be performed normally exceeds a predetermined period in the protocol. And the connection route is switched when the hold timer expires.

A method for reducing the processing time required for the BGP scan is disclosed in Patent Document 1. That is, the means for solving the problem is that the address of the PE router that is the next hop distributed by the BGP protocol and the address of the C (core) router that is the next hop distributed by the OSPF (Open Shortest Path First) protocol are the next hop. This is a means for registering in association with a database, detecting a changed next hop by scanning of a scan processing unit, and a BGP database changing unit changing a reliable next hop of only changed route information.
JP 2005-20604 A

  In the above-described conventional configuration, one router has a route time other than the route information distributed from the other router. (The default value of BGP is 180 seconds), and there is a problem that switching is delayed.

  Accordingly, an object of the present invention is to provide a connection management device, a connection management device control method, and a control program capable of quickly switching to a bypass connection route when an abnormality is detected in the connection route. .

In order to solve the above-described problem, a connection management apparatus that exchanges route information according to a connection communication type route information exchange protocol between each network when connecting a plurality of networks, and BGP between other networks In a connection management device that establishes a peer and performs communication , a connection route is secured for another connection device on which the BGP peer is established on the first connection route managed by the connection management device. An inquiry unit that makes an inquiry by transmitting an ICMP echo packet via a predetermined transmission interface unit, and a route that is different from the first connection route and is connected to the other connection device ICMP echo packet that another connection management device that manages the second connection path performs to the other connection device via the second connection path The path information is configured so that the priority of the first connection path is relatively higher than that of the second connection path when it is detected that the response by the ICMP echo reply packet corresponding to the inquiry by the user cannot be received. And a path attribute changing unit that changes a path attribute to be performed.

The connection device is a router, an L2 switch, or an L3 switch, and the connection management device is configured as a router or a route reflector.

A connection management device for exchanging route information according to a connection communication type route information exchange protocol between multiple networks when establishing a BGP peer with other connected devices. In the connection management device control method to be performed, whether or not a connection route is secured for another connected device on which the BGP peer is established on the first connection route managed by the connection management device. An inquiry process for making an inquiry by transmitting an ICMP echo packet via a predetermined transmission interface unit, and a second connection that is different from the first connection path and is connected to the other connection device Inquiry based on ICMP echo packet made by other connection management device managing the route to the other connected device via the second connection route The path information is configured to raise the priority of the first connection path relative to the second connection path when it is detected that the response by the ICMP echo reply packet corresponding to the call cannot be received. And a path attribute changing process for changing the path attribute.

The inquiry unit transmits an ICMP echo packet, and the path attribute changing unit changes the path attribute when the other connection management device cannot receive an ICMP echo reply packet for the ICMP echo packet. It is a feature.

In the connection management apparatus control method for exchanging path information between networks according to a connection communication type path information exchange protocol when connecting a plurality of networks, the first connection managed by the connection management apparatus An inquiry process for making an inquiry about whether or not a connection route is secured to other connection devices on the route via a predetermined transmission interface unit, and a route different from the first connection route, Another connection management device that manages the second connection path connected to the other connection device cannot receive a response corresponding to the inquiry made to the other connection device via the second connection path. A path attribute that changes the path attribute that constitutes the path information so that the priority of the first connection path is relatively higher than that of the second connection path. Is characterized by comprising a changing process, the.

  According to the present invention, when an abnormality is detected in a connection path, it is possible to quickly switch to a detour connection path, thereby improving throughput.

  Next, a preferred embodiment of the present invention will be described.

  The first router (connection management device) periodically repeats the transmission of an ICMP echo packet (ping) to the second router (connection management device or connection device) that is a BGP peer. When an abnormality occurs, the ICMP echo reply packet from the second router cannot be received.

Therefore, in the first router, it is confirmed whether or not a connection path with the second router connection management device is secured by transmitting the ICMP echo packet / receiving the ICMP echo reply packet. When an abnormality is detected, one of the following processes (1) to (3) is performed.
(1) Interface invalidation processing (2) Connection route invalidation processing (3) Connection route redistribution processing

  First, the interface invalidation processing will be described.

FIG. 1 is a process flowchart of an interface invalidation process.
The first router (connection management device) sends an ICMP echo packet (ICMP echo message) to a second router (another connection management device or connection device) that is BGP-connected via a predetermined transmission interface for a predetermined time. It transmits every time (step S11).

  Abnormality detection is performed based on whether or not the corresponding ICMP echo reply packet (ICMP echo reply message) is received after the ICMP echo packet is transmitted (step S12), and the transmission interface corresponding to the connection path in which the abnormality is detected is invalidated. (Step S13).

  Here, the invalidation of the transmission interface does not release the physical link, but performs an operation that makes the logical interface (interface recognized by each protocol) appear to be invalidated.

  For this reason, even when one physical link is used in a plurality of virtual LANs, control is possible for each virtual LAN.

  Next, connection path invalidation processing will be described.

FIG. 2 is a process flowchart of the connection path invalidation process.
When a BGP session is configured between the first router (connection management device) and the second router (other connection management device or connection device), the first router (connection management device) has a predetermined transmission interface. An ICMP echo packet is transmitted at predetermined time intervals to a second router (another connection management device or connection device) that is connected via BGP to the peer (step S21).

  The first router detects an abnormality in the reception of the ICMP echo reply packet with respect to the transmission of the ICMP echo packet (step S22), and when an abnormality is detected, invalidates all the routes distributed from the second router. Then, an operation of withdrawing the connection route distributed from the second router is performed on another connection device connected to another BGP peer (step S23).

  As described above, the connection route distributed from the second router is withdrawn, so that the priority order of the other connection routes is increased and the best connection route is selected.

  Next, connection path redistribution processing will be described.

FIG. 3 is a process flowchart of the connection path redistribution process.
It is assumed that the third router forms a BGP session with the first router (first connection management device) and the second router (second connection management device).

  In the third router, certain route information is distributed from both the first router and the second router, and normally, the route information from the first router is selected as the best connection route. Suppose there is.

  In this case, if the connection route between the first router and the third router becomes abnormal, the third router cannot connect the route based on the route information distributed from the first router. .

  However, since the third router cannot detect the line abnormality, the route information is not switched.

  Therefore, the second router monitors the BGP session between the first router and the third router, that is, monitors that the first router has transmitted an ICMP echo packet (step S31). ), Monitoring whether the corresponding ICMP echo reply packet is received (step S32), and when reachability to the second router becomes abnormal, between the third router and the second router In the BGP session, the path attribute is changed (step S33), and the route information corresponding to the route information distributed by the first router to the third router is redistributed.

  When redistributing this route information, the third router performs processing for changing a part of the attribute information of the route information in the third router based on the route information distributed from the first router. This is because the priority is given to the route information distributed from.

  As a result, the third router switches to the route information earlier than the BGP session with the first router is disconnected by the protocol timer.

  Since the second router changes the attribute information of the route information and redistributes the third router, the route passing through the second router has the highest priority by comparing the routes defined in the protocol. It can be selected as a thing. That is, an existing BGP router can be used as the third router.

Next, about specific embodiment,
(1) Interface Invalidation Processing (2) Connection Route Invalidation Processing (3) Connection route redistribution (update) processing will be described.

[1] First Embodiment The first embodiment is an embodiment in the case of performing an interface invalidation process.
FIG. 4 is a schematic configuration diagram of a network according to the first embodiment.
The network 10 constitutes a provider network 11, a first customer network, a customer edge router 13 to which a user terminal 12 is connected, a customer edge router 14 that constitutes a second network, the provider network 11, and a customer And an L2 switch 15 for connecting to the edge router 13.

  The provider network 11 includes a provider edge router 16 connected to the customer edge router 13 via the L2 switch 15, a provider edge router 17 connected to the customer edge router 14, and a provider edge router connected to the customer edge router 13. 18.

  Next, the configuration of the router will be described. Since the routers 13, 14, 16, 17, and 18 have substantially the same configuration, the provider edge router 16 will be described as an example.

FIG. 5 is a schematic block diagram of the provider edge router.
The provider edge router 16 includes a kernel unit 16A that controls the entire provider edge router 16, and an ICMP monitoring processing unit that transmits an ICMP echo packet and receives an ICMP echo reply packet to monitor whether the connection path is normal. 16B, a BGP protocol processing unit 16D that performs BGP protocol processing with reference to the BGP route information database (DB) 16C under the control of the kernel unit 16A, and performs communication according to the BGP protocol, and under the control of the kernel unit 16A And a route information processing unit 16F that performs processing of route information related to the actual connection route with reference to the route information database (DB) 16E.

  Here, the kernel unit 16A is realized as a microcomputer including a CPU, a ROM, a RAM, and the like, and performs various control processes to be described later based on a control program.

  With this configuration, the ICMP monitoring processing unit 16B transmits an ICMP echo packet and receives an ICMP echo reply packet to monitor whether the connection path is normal.

  If the connection path is normal as a result of this monitoring, the path information processing section 16F refers to the path information database (DB) 16E and processes path information regarding the actual connection path under the control of the kernel section 16A. The BGP protocol processing unit 16D performs processing of the BGP protocol with reference to the BGP route information database (DB) 16C, and communicates with other target connection devices according to the BGP protocol.

  Further, when the connection path is abnormal, the ICMP monitoring processing unit 16B notifies the kernel unit 16A and the BGP protocol processing unit 16D that the abnormality has occurred, and also notifies when the recovery is performed. As a result, in addition to the communication process, the BGP protocol processing unit 16D performs at least one of the above-described interface invalidation process, connection path invalidation process, or connection path redistribution (update) process.

  Next, the outline operation will be described.

First, the operation when the interface invalidation process is performed will be described.
In this case, the provider edge router 16 and the customer edge router 13 have established a BGP peer.

  In a state where the BGP peer is established, the provider edge router 16 periodically transmits an ICMP echo packet to the customer edge router 13 via a predetermined interface and the L2 switch 15.

  The customer edge router 13 that has received the ICMP echo packet transmits an ICMP echo reply packet to the provider edge router 16 via the L2 switch 15 in order to maintain the establishment of the BGP peer.

  As a result, when the provider edge router 16 receives the ICPM echo reply packet from the customer edge router 13, when it reaches the customer edge router 13, it assumes that the connection path is normal and continues normal processing.

  On the other hand, when the connection path between the L2 switch 15 and the customer edge router 13 becomes abnormal, the provider edge router 16 cannot receive the ICMP echo reply packet from the customer edge router 13.

  Therefore, the provider edge router 16 transmits the ICMP echo packet to each device constituting the network when it cannot receive the ICMP echo reply packet within a predetermined time after transmitting the ICMP echo packet. It appears that the specified interface is down.

  As a result, a static (fixed) route that uses the interface as a transmission interface can be invalidated, and a protocol that configures a session on the interface can be down.

  As a result, the connection route through the interface becomes invalid, and a connection route different from the connection route is immediately selected as the best connection route (best path), and there is no time lag and the connection route Switching can be performed.

  In this case, making the interface appear to be down is an operation that makes the logical interface recognized by each protocol appear as if it is down without dropping the physical link.

  Therefore, even when one physical interface is shared by a plurality of virtual LANs (VLANs), it is possible to control each virtual LAN.

[2] Second Embodiment The second embodiment is an embodiment in the case of performing a connection path invalidation process.

FIG. 6 is a schematic configuration diagram of a network according to the second embodiment.
The network 20 constitutes a provider network 21 and a first customer network, a customer edge router 22 to which a user terminal (not shown) is connected, and a second network and a user terminal (not shown) are connected. And a customer edge router 23.

  The provider network 21 includes a provider edge router 25 connected to the customer edge router 22, a provider edge router 26 similarly connected to the customer edge router 22, a provider edge router 27 connected to the customer edge router 23, The route reflector 28 is connected to the provider edge routers 25 to 27, collects route information from the provider edge routers 25 to 27, and redistributes the route information.

Next, the operation when the connection path invalidation process is performed will be described.
In this case, the route reflector 28 and the provider edge routers 25 to 27 each establish a BGP peer (iBGP peer: internal BGP peer).

  Here, a case where a data transmission operation is performed from the customer edge router 23 side to the customer edge router 26 side via the provider edge router 27 will be described.

  In a state where the BGP peer is established, the route reflector 28 periodically transmits ICMP echo packets to the provider edge routers 25 to 27 via a predetermined interface.

  If the route reflector 28 does not receive the ICMP echo reply packet within a predetermined time after transmitting the ICMP echo packet, the connection collected from the provider edge router 25 that has not received the ICMP echo reply packet. All the route information is invalidated, and the connection route information distributed and collected from the provider edge router 25 is withdrawn to other provider edge routers 26 and 27 connected to other BGP peers.

  Specifically, when the terminal device A is connected to the customer edge router 22 as connection path information in a state in which the ICMP echo reply packet is normally received in the route reflector 28, for example, the provider edge router 25 is represented as PE1, and the provider edge router 26 is represented as PE2.

At this time, from the provider edge router 25 as connection path information,
A nexthop PE1
Is collected by the route reflector 28. Similarly, from the provider edge router 26 as connection path information,
A nexthop PE2
Is collected by the route reflector 28.
Then, from the route reflector 28 to the provider edge router 27 as connection path information,
A nexthop PE1
A nexthop PE2
Will be redistributed.

Thereafter, when the route reflector 28 cannot normally receive the ICMP echo reply packet from the provider edge router 25, it is connection path information distributed from the provider edge router 25.
A nexthop PE1
Is withdrawn, which is connection path information via the provider edge router 26.
A nexthop PE2
Is selected as the best connection route, and the data from the customer edge router 23 side is transmitted to the customer edge router 22 side via the provider edge router 27 and the provider edge router 26.

[3] Third Embodiment The third embodiment is an embodiment in the case of performing redistribution processing of a connection route. The difference between the third embodiment and the second embodiment described above is that, in the second embodiment, the connection route information corresponding to the connection route in which an abnormality is recognized is withdrawn so as to be invalidated. In the form, the connection route information corresponding to the connection route in which an abnormality is recognized is not withdrawn, and the priority is lowered.

FIG. 7 is a schematic configuration diagram of a network according to the third embodiment.
The network 30 includes a provider network 31 and a customer edge router 32 that constitutes a customer network and to which the user terminal device A is connected.

  The provider network 31 is connected to the provider edge router 33 connected to the customer edge router 32, the provider edge router 34 connected to the customer edge router 32, and the provider edge routers 33 and 34. The first route reflector 35 that collects and redistributes route information from 33 and 34 is connected to each provider edge router 33 and 34 in the same manner, collects route information from each provider edge router 33 and 34, and redistributes it. And a second route reflector 36.

Next, the operation when the connection path redistribution processing is performed will be described.
In this case, the route reflector 35 and the route reflector 36 and the provider edge routers 33 and 34 each establish a BGP peer (iBGP peer: internal BGP peer).

  Here, a case where the user terminal (A) 37 performs a data transmission operation to the user terminal (B) 38 from the customer edge router 32 side will be described.

  In the state where the BGP peer is established, the route reflectors 35 and 36 periodically send ICMP echo packets to the provider edge routers 33 and 34 through predetermined interfaces, respectively.

  The route reflector 35 and the route reflector 36 monitor the BGP session of the other reflector reflector, the other route reflector transmits an ICMP echo packet, and the corresponding ICMP echo reply packet is transmitted within the predetermined time to the other route reflector. If the route reflector that has received the ICMP echo reply packet cannot receive the ICMP echo reply packet, it receives the ICMP echo reply packet from the provider edge router corresponding to the ICMP echo packet that cannot be received by the other route reflector. New connection route information having a higher priority than the priority of the connection route information distributed by the route reflector that could not be generated is generated and redistributed.

  Here, it demonstrates more concretely. Here, the provider edge router 33 is represented as PE1, the provider edge router 34 as PE2, the route reflector 35 as RR1, and the route reflector 36 as RR2.

For example, when the terminal device B is connected to the provider edge router 34 as connection path information in a state in which the ICMP echo reply packet is normally received in the route reflector 35, the route reflector 35 to the customer edge router 32. For connection route information,
B next PE2 (RR1) MED = 100
(B here means the user terminal 38). Similarly, from the route reflector 36, as connection path information,
B next PE2 (RR2) MED = 200
Is distributed. Here, the customer edge router 32 refers to the MED (Multi Exit Discriminator) value, and the connection corresponding to the connection route information distributed from the route reflector 35 having a smaller MED value (higher priority). The route will be selected.

  In this state, if an abnormality occurs in the connection path between the route reflector 35 and the provider edge router 33, the customer edge router 32 may normally transmit using the connection path information distributed to the route reflector 35. become unable.

  However, the customer edge router 32 cannot detect an abnormality in the connection route via the provider edge router 33, and a predetermined time (for example, 180 seconds) corresponding to the hold timer value defined by the protocol. Until the time elapses, the connection path cannot be switched.

By the way, in this state, the route reflector 35 cannot receive the ICMP echo reply packet normally. On the other hand, the route reflector 36 monitors the BGP session of the route reflector 35 and detects that a data transmission abnormality has occurred with respect to the customer edge router 32 of the route reflector 35, for example, as connection path information,
B nexthop PE2 (RR2) MED = 50
Are redistributed to the customer edge router 32.

  As a result, the customer edge router 32 refers to the MED (Multi Exit Discriminator) value, and the connection edge information distributed from both the route reflector 35 and the route reflector 36 has a smaller MED value (more The connection route corresponding to the connection route information distributed from the route reflector 36 is selected, and the route to the user terminal B can be switched earlier than the BGP session with the route reflector 35 is disconnected. .

  As described above, according to the third embodiment, a customer edge router having a conventional BGP function can be used as the customer edge router 32, and the route reflector of the present invention is applied to an existing network. By doing so, it becomes possible to easily improve the function of the network.

  In the above description, as a method for increasing the priority of the connection route information, the case where the value of MED as a path attribute is changed has been described. However, the path attribute to be changed is not limited to MED but WEIGHT (value is It is possible to use LOCAL_PREFRENCE (a larger value is a higher priority), router-ID (a smaller value is a higher priority), etc.

[4] Effects of Embodiments As described above, according to each embodiment, even with a connection device (router) that cannot directly detect an abnormality in a connection route, a route by a hold timer defined by the protocol It is possible to perform the same operation as when a connection path abnormality is detected at a timing earlier than the switching timing, and it is possible to switch to a detour connection path at high speed, thereby improving throughput.

[5] Modification of Embodiment In the above description, the network configuration is an example, and connection management devices (route reflectors, routers, etc.) that have established BGP peers or between the connection management device and the connection device (router, L2 switch) , L3 switch, etc.), the present invention can be applied.

It is a process flowchart of the invalidation process of an interface. It is a process flowchart of the invalidation process of a connection path | route. It is a process flowchart of the redistribution process of a connection path | route. It is a schematic block diagram of the network of 1st Embodiment. It is a general | schematic block diagram of a provider edge router. It is a schematic block diagram of the network of 2nd Embodiment. It is a schematic block diagram of the network of 3rd Embodiment.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... Network, 11 ... Provider network, 12 ... User terminal, 13 ... Customer edge router, 14 ... Customer edge router, 15 ... L2 switch, 16 ... Provider edge router, 16A ... Kernel part, 16B ... ICMP monitoring processing part, 16C ... BGP route information database (DB), 16D ... BGP protocol processing unit, 16E ... Route information database (DB), 16F ... Route information processing unit, 17 ... Provider edge router, 18 ... Provider edge router, 20 ... Network, 21 ... Provider network, 22 ... customer edge router, 23 ... customer edge router, 25 ... provider edge router, 26 ... provider edge router, 27 ... provider edge router, 28 ... route reflector, 30 ... network, 31 ... B provider network, 32 ... customer edge router, 33 ... provider edge router, 34 ... provider edge router, 35 ... first route reflector, 36 ... second route reflector.

Claims (3)

A connection management device for exchanging route information according to a connection communication type route information exchange protocol between multiple networks when establishing a BGP peer with other connected devices. In the connection management device to perform
Whether or not a connection path is secured to another connection device on which the BGP peer is established on the first connection path managed by the connection management apparatus is determined via ICMP through a predetermined transmission interface unit. An inquiry unit that makes an inquiry by sending an echo packet ;
Another connection management device that is different from the first connection path and manages the second connection path connected to the other connection device is configured to transmit the second connection path via the second connection path. When it is detected that the response by the ICMP echo reply packet corresponding to the inquiry by the ICMP echo packet made to another connected device cannot be received, the priority of the first connection route is relatively set to the second connection route. A path attribute changing unit that changes a path attribute that constitutes the route information;
A connection management device comprising: The connection management device according to claim 1,
The connection device is a router, an L2 switch, or an L3 switch,
The connection management device is configured as a router or a route reflector. A connection management device for exchanging route information between networks according to a connection communication type route information exchange protocol when connecting a plurality of networks, and establishing a BGP peer with other connected devices for communication In the connection management device control method to be performed,
Whether or not a connection path is secured to another connection device on which the BGP peer is established on the first connection path managed by the connection management apparatus is determined via ICMP through a predetermined transmission interface unit. An inquiry process for making an inquiry by sending an echo packet;
Another connection management device that is different from the first connection path and manages the second connection path connected to the other connection device is configured to transmit the second connection path via the second connection path. When it is detected that the response by the ICMP echo reply packet corresponding to the inquiry by the ICMP echo packet made to another connected device cannot be received, the priority of the first connection route is relatively set to the second connection route. Path attribute changing process for changing the path attribute that constitutes the route information,
A method for controlling a connection management apparatus comprising:

Images