JP5558436B2 - Network system and network failure avoidance method - Google Patents

Description

  The present invention is configured by an L2 switch (layer 2 switch) that accommodates user terminals and an L3 switch (layer 3 switch) that bundles L2 switches, and is connected to the L3 switch in a network system having a redundant configuration using link aggregation. The present invention relates to a network system and a network failure avoidance method for improving the reliability of communication between user terminals against simultaneous failure of a plurality of lines.

  The physical configuration, logical configuration, and user communication path of the network system as the background of the present invention will be described with reference to FIG.

(Physical configuration of network system)
In FIG. 5, the user terminal 11 is connected to the L2 switch 21, the user terminal 12 is connected to the L2 switch 22, and the user terminal 13 is connected to the L2 switch 23. Here, for simplicity, one user terminal is connected to one L2 switch, but a plurality of user terminals may be connected to the L2 switch. The L2 switch 21 is connected to both the L3 switch 31 and the L3 switch 32 via the line 1 and the line 2. Similarly, the L2 switch 22 is connected to both the L3 switch 31 and the L3 switch 32 via the line 3 and the line 4, and the L2 switch 23 is connected to the L3 switch 31 and the L3 switch 32 via the line 5 and the line 6. Connected to both. The line 7 is a line for connecting the L3 switch 31 and the L3 switch 32. Even if the line 7 is directly connected between the L3 switch 31 and the L3 switch 32, the line 7 is connected via a network device such as a router. And a virtual line connecting the L3 switch 32 may be used.

(Logical configuration of network system)
As viewed from the L2 switch 21, the L3 switch 31 and the L3 switch 32 are two devices connected via the line 1 and the line 2 which are physically different lines, but the link aggregation 41 (non-patent document) By the apparatus redundant connection technology using 1), each logically operates as one line and one apparatus. Similarly, by configuring the link aggregation 42 in which the lines 3 and 4 are bundled, the L3 switch 31 and the L3 switch 32 are logically recognized as one L3 switch as viewed from the L2 switch 22. Further, by configuring the link aggregation 43 in which the lines 5 and 6 are bundled, the L3 switch 31 and the L3 switch 32 are logically recognized as one L3 switch as viewed from the L2 switch 23.

(User communication path of network system)
On the link aggregations 41, 42, and 43, load distribution of the communication band is performed by load balancing. The output destination line of the transfer frame in the load balance is determined by the internal processing of each of the L2 switches 21, 22, and 23 based on an arbitrary transfer frame distribution rule such as a flow unit or a destination unit. When the load balance is performed, the communication path from the user terminal 11 to the user terminal 12 includes two patterns of the user communication path 51 and the user communication path 52 shown in FIG. 5 and two patterns (not shown).

(1) User communication path 51: user terminal 11 → L2 switch 21 → line 1 → L3 switch 31 → line 3 → L2 switch 22 → user terminal 12
(2) User communication path 52: user terminal 11 → L2 switch 21 → line 2 → L3 switch 32 → line 4 → L2 switch 22 → user terminal 12

(3) User communication path 51 ′ (not shown): user terminal 11 → L2 switch 21 → line 1 → L3 switch 31 → line 7 → L3 switch 32 → line 4 → L2 switch 22 → user terminal 12
(4) User communication path 52 ′ (not shown): user terminal 11 → L2 switch 21 → line 2 → L3 switch 32 → line 7 → L3 switch 31 → line 3 → L2 switch 22 → user terminal 12

IEEE802.1AX IEEE Standard for Local and metropolitan area networks Link Aggregation Request for Comments: 792 Internet Control Message Protocol Request for Comments: 1442 Structure of Management Information for version 2 of the Simple Network Management Protocol

(Mechanism of simultaneous failure of multiple lines and inability to communicate with users)
In the network system of FIG. 5, a mechanism that disables communication between users due to simultaneous failure of a plurality of lines connected to the L3 switch will be described with reference to FIG.

  In FIG. 6, when the L3 switch 31 and the L3 switch 32 are connected to the L2 switch 21, for example, if a plurality of lines 3 and lines 7 ahead of the one L3 switch 31 fail simultaneously, the L3 switch 31 to the L2 switch 22. The transfer frame for will not be delivered. That is, since the L2 switch 21 cannot recognize that the reachability of the transfer frame from the L3 switch 31 has been lost, in the communication from the user terminal 11 to the user terminal 12, based on the transfer frame distribution rule of the link aggregation 41, When the L2 switch 21 transfers a part of the transfer frame to the L3 switch 31, the transferred frame is discarded as a destination unknown packet in the L3 switch 31.

(Problem solving method and its harmful effects in the prior art)
For the simultaneous failure of a plurality of lines as shown in FIG. 6, communication failure at the time of the simultaneous failure of a plurality of lines can be avoided by performing apparatus redundancy by dynamic routing. However, in order to perform dynamic routing setting between network devices, it is necessary to replace the L2 switch with an L3 switch, and there is a concern that the operability will be degraded, such as complicated network design and increased required device settings. Further, since the L3 switch is generally expensive compared to the L2 switch, there is a problem that the introduction cost is higher than that of the network configured by the L2 switch.

  In the network system in which the L2 switch and the L3 switch are connected redundantly by using link aggregation, the present invention has a reliability for avoiding packet loss at the time of simultaneous failure of a plurality of lines. An object is to provide a high network system and a network failure avoidance method.

A first invention includes a plurality of L2 switches to which one or more user terminals are connected, and a plurality of L3 switches to which a plurality of L2 switches are connected, respectively, and each of the plurality of L2 switches and a plurality of L3 switches In the network system in which the lines between the two are bundled by link aggregation and a plurality of L3 switches are connected to each other, the L3 switch detects a failure of the line directly connected to itself, and a plurality of L3 switches Means for transmitting a monitoring packet destined for a predetermined L3 switch and detecting a failure of the virtual circuit according to the presence or absence of reception of the response signal when the line connecting the switches to each other is a virtual line; All of the redundant lines and virtual lines that transfer frames to a given L2 switch Means for notifying the L2 switch other than the predetermined L2 switch of the failure, and the L2 switch has the user terminal connected to the predetermined L2 switch as the destination by the notification of the failure. The transfer frame to be transmitted is excluded from the transfer frame distribution process at the time of link aggregation load balancing, and includes a means for setting to transmit only from the line connected to the L3 switch that does not notify the failure.

In the network system according to the first aspect of the invention, the means for notifying a failure from the L3 switch uses a management protocol to notify the failure state, or uses a remote connection protocol without notifying the failure itself, and other than a predetermined L2 switch. The configuration is changed to switch the line for transferring the frame to the L2 switch.

In the network system of the first invention, the means for notifying the failure of the L3 switch is configured to notify an L2 switch other than the predetermined L2 switch via a terminal that manages network information.

A second invention includes a plurality of L2 switches to which one or more user terminals are connected, and a plurality of L3 switches to which a plurality of L2 switches are connected, respectively, and each of the plurality of L2 switches and a plurality of L3 switches In the network failure avoiding method of a network system in which a line between the two is bundled by link aggregation and a plurality of L3 switches are connected to each other, the L3 switch detects a failure of a line directly connected to itself. In addition, when a line connecting a plurality of L3 switches to each other is a virtual line, a monitoring packet destined for a predetermined L3 switch is transmitted, and a failure of the virtual line is detected depending on whether or not a response signal is received. A redundant line for transferring a frame to a predetermined L2 switch among the plurality of L2 switches and a temporary line. When all of the lines are detected as having failed, the failure is notified to L2 switches other than the predetermined L2 switch, and the L2 switch is connected to the predetermined L2 switch by the notification of the failure. Is set to be transmitted only from the line connected to the L3 switch that does not notify the failure, excluding the transfer frame allocation processing at the time of link aggregation load balancing.

In the network failure avoidance method according to the second aspect of the present invention, the failure notification from the L3 switch uses the management protocol to notify the failure state, or uses the remote connection protocol without notifying the failure itself, and other than the predetermined L2 switch. The setting is changed to switch the line for transferring the frame to the L2 switch.

  In the network failure avoidance method according to the second aspect of the invention, the failure notification from the L3 switch is notified to an L2 switch other than the predetermined L2 switch via a terminal that manages network information.

  In the present invention, in a network system in which L2 switches are redundantly connected via link aggregation to a redundant L3 switch, the L2 switch detects a failure pattern in which communication between users is disabled, and the failure pattern By switching the output destination line triggered by the detection, user communication can be continued even when a plurality of lines fail simultaneously. Since L2 switch functions can be expanded by software extension, avoiding a decrease in operability due to inability to communicate while using the L2 switch, and constructing a network system at a lower cost than introducing a new L3 switch Can do.

It is a figure which shows the example of a detection process of the failure pattern by the autonomous detection system of Example 1 of this invention. It is a figure which shows the example of a failure avoidance process at the time of the failure detection by the autonomous detection system of Example 1 of this invention. It is a figure which shows the example of a detection process of the failure pattern by the failure notification system of Example 2 of this invention. It is a figure which shows the example of a failure avoidance process at the time of the failure detection by the failure notification system of Example 2 of this invention. It is a figure explaining the physical structure of a network system, a logical structure, and a user communication path | route. It is a figure explaining the mechanism in which communication between users becomes impossible.

  In the present invention, in order to avoid inability to communicate between users, the L2 switch first detects a failure pattern in which communication between users is disabled, and loses reachability to the user terminal when the failure pattern is detected. The frame transfer to the L3 switch is stopped, and the frame transfer destination is switched to the L3 switch that has not lost reachability. As the failure pattern detection method, in the present invention, the L2 switch detects the failure pattern by itself, and the failure notification method in which another device such as the L3 switch detects the failure pattern and notifies the L2 switch. Exists. Each method will be described below as a first embodiment and a second embodiment.

  FIG. 1 shows an example of failure pattern detection processing by the autonomous detection method according to the first embodiment of the present invention. Note that the physical configuration, logical configuration, and user communication path of the network system are the same as those of the network system shown in FIG.

  In FIG. 1, the L2 switch 21 checks the frame reachability from the L3 switch 31 to the L2 switch 22 and the frame reachability from the L3 switch 22 to the L2 switch 22, so that the transmission source is the L2 switch 21 and the destination is L2 A probe packet for the switch 22 is transmitted. Since the reachability monitoring across the L3 switches is performed as the probe packet, for example, an L3 monitoring packet such as ICMP (Internet Control Message Protocol) shown in Non-Patent Document 2 is used.

  A normal L2 switch does not distinguish between physical lines constituting link aggregation, and an L3 monitoring packet is transmitted from either the line 1 or the line 2 by load balancing. However, in the L2 switch 21 according to the present invention, the line 1 And the line 2 are transmitted with the L3 monitoring packet having the transmission source as the L2 switch 21 and the destination as the L2 switch 22. Here, the L2 switch 21 separately manages the identifier information added to the L3 monitoring packet transmitted from the line 1 and the line 2. For example, when ICMP Ping is used as the L3 monitoring packet, an identifier field in the ICMP field is used.

  If no failure has occurred, the L3 monitoring packet reaches the L2 switch 22 via the communication path 61 and the communication path 62. The L2 switch 22 receives the L3 monitoring packet transferred via the communication path 61 and the communication path 62, generates a response packet including the identifier information thereof, and sets the transmission source as the L2 switch 22 and the destination as the L2 switch. 21 is transmitted. The L2 switch 21 receives the response packet from the L2 switch 22 and confirms each identifier, thereby confirming that there is frame reachability from the L3 switch 31 and the L3 switch 32 to the L2 switch 22. The transmission operation of the L3 monitoring packet in the L2 switch 21 may be performed periodically at an arbitrary time interval, or may be manually performed by a network operator.

  Here, when there is a response to only one of the L3 monitoring packets transmitted from the lines 1 and 2, the L2 switch 21 receives the identifier information of the L3 monitoring packets transmitted from the lines 1 and 2, and the L2 switch 22 By associating with the identifier information of the response packet from, it is possible to grasp which line is the response to the L3 monitoring packet transmitted from. For example, if the reachability of the transfer frame from the L3 switch 31 to the L2 switch 22 is lost due to simultaneous failure of the line 3 and the line 7, a response packet to the L3 monitoring packet transmitted from the line 1 cannot be received. . If the reachability of the transfer frame from the L3 switch 32 to the L2 switch 22 is lost due to the simultaneous failure of the lines 4 and 7, the response packet for the L3 monitoring packet transmitted from the line 2 cannot be received. . As described above, when the L2 switch 21 responds to only one of the L3 monitoring packets transmitted from the lines 1 and 2, the L2 switch 21 detects that a failure pattern that disables communication between users has occurred. can do.

  In the first embodiment, the destination of the L3 monitoring packet from the L2 switch 21 is the L2 switch 22, but the user terminal 12 may be the destination. When the destination of the L3 monitoring packet is the user terminal 12, the source of the response packet for the L3 monitoring packet is the user terminal 12.

  When the L2 switch 21 detects occurrence of a failure pattern that disables communication between users in this way, the frame transfer destination switching operation is performed so as to avoid the failure path. Note that in the normal L2 switch, since the transfer frame distribution for determining the output line by load balance is executed by the internal processing of the apparatus, the output destination line of a specific frame cannot be explicitly specified.

  The L2 switch according to the present invention is provided with a function capable of explicitly designating from which line a transfer frame having designated destination information is excluded from a transfer frame distribution process at the time of load balancing. For example, when the L2 switch 22 or the user terminal 12 is the destination and there is no response to the L3 monitoring packet transmitted from the line 1, the transfer frame destined for the user terminal 12 is changed to the transfer frame distribution rule at the time of load balancing. The correspondence relationship of excluding the target is set in the L2 switch 21 as the preset information.

  FIG. 2 shows an example of failure avoidance processing when a failure is detected by the autonomous detection method according to the first embodiment of the present invention. In FIG. 2, when the L2 switch 21 cannot obtain a response packet to the L3 monitoring packet transmitted from the line 1 with respect to the L3 monitoring packet transmitted from the line 1 and the line 2, the L3 switch 21 and the line 3 connected to the L3 switch 31 It is determined that a failure has occurred in both lines 7. The L2 switch 21 having made this determination excludes the transfer frame destined for the user terminal 12 from the transfer frame distribution process at the time of load balancing, and transmits the transfer frame only from the line 2 that has received the response packet. Communication from the terminal 11 to the user terminal 12 is transferred only through the user communication path 52.

  Similarly, in the L2 switch 21, when the path from the L3 switch 31 and the L3 switch 32 to the L2 switch 23 is confirmed to be normal, the frame destined for the user terminal 13 connected to the L2 switch 23 is loaded. It is a target of transfer frame distribution processing at the time of balance, and is transmitted from the line 1 or 2 to either the L3 switch 31 or the L3 switch 32. That is, frames other than those destined for the user terminal 12 are subject to transfer frame distribution processing at the time of load balancing, and are transmitted from either the line 1 or the line 2.

  FIG. 3 shows an example of failure pattern detection processing by the failure notification method according to the second embodiment of the present invention. Note that the physical configuration, logical configuration, and user communication path of the network are the same as those of the network shown in FIG.

  In FIG. 3, in the failure notification method, first, the reachability monitoring from the L3 switch itself to the L2 switch is performed in the L3 switch. In order to confirm the frame reachability from the L3 switch 31 to the L2 switch 22, it is necessary to monitor the line 3 and the line 7. Since the line 3 is a directly connected line when viewed from the L3 switch 31, it is possible to monitor the presence or absence of a failure on the line 3 without using a probe packet. Similarly, when the line 7 is a physical line and the L3 switch 31 and the L3 switch 32 are directly connected via the line 7, the L3 switch 31 determines whether there is a failure on the line 7 without using a probe packet. Can be monitored. On the other hand, when the line 7 is a virtual line and the L3 switch 31 and the L3 switch 32 are connected via a plurality of network devices such as switches and routers, the frame reachability from the L3 switch 31 to the L3 switch 32 Is required to transmit a probe packet having the transmission source as the L3 switch 31 and the destination as the L3 switch 32. As the probe packet, the same L3 monitoring packet as in the first embodiment and a response packet for the same can be used. The probe packet transmission operation may be performed periodically at an arbitrary time interval, or may be manually performed by a network operator.

  Similarly, the transfer frame reachability from the L3 switch 31 to the L2 switch 23 is confirmed by the L3 switch 31 monitoring the line 5 and the line 7. Further, the transfer frame reachability from the L3 switch 32 to the L2 switch 22 is confirmed by the L3 switch 32 monitoring the line 4 and the line 7. The transfer frame reachability from the L3 switch 32 to the L2 switch 23 is confirmed by the L3 switch 32 monitoring the line 6 and the line 7.

  Here, when failures occur in the line 3 and the line 7 at the same time, the reachability of the transfer frame from the L3 switch 31 to the L2 switch 22 is lost, so the L3 switch 31 has a failure pattern in which communication between users becomes impossible. It is determined that there is, and notifies the L2 switch 21 and the L2 switch 23 that the failure pattern has occurred. There are two patterns for notifying failure.

  The first pattern is a method of notifying the failure state of the line 3 and the line 7 using a management protocol such as SNMP (Simple Network Management Protocol) shown in Non-Patent Document 3, for example. In the second pattern, there is a method of changing the setting for switching the frame transfer destination in the L2 switch 21 using a remote connection protocol such as TELNET without notifying the failure itself. The failure notification path from the L3 switch 31 to the L2 switch 21 may be notified directly from the L3 switch 31 to the L2 switch 21 as in the notification path 71, or in a terminal that manages network information as in the notification path 72. You may notify via a certain network management terminal 81. FIG.

  FIG. 4 shows an example of failure avoidance processing when a failure is detected by the failure notification method according to the second embodiment of the present invention. In FIG. 4, the L2 switch 21 has a function of explicitly selecting an output line by excluding a transfer frame having designated destination information from a transfer frame distribution process during load balancing, as in the first embodiment. . The L2 switch 21 transmits a frame destined for the user terminal 12 only from the line 2 in response to a failure notification from the L3 switch 32 or the network management terminal 81, and communication from the user terminal 11 to the user terminal 12 is performed by the user. It is transferred only through the communication path 52.

  Frames whose destination information is other than the user terminal 12 are subjected to transfer frame distribution processing during load balancing, and are transmitted from either the line 1 or the line 2. When a failure occurs in the line 3 and the line 7, the correspondence relationship that the transfer frame destined for the user terminal 12 is transmitted from the line 2 is set in the L2 switch 21 in advance, or the L3 switch 31 or the network It may be set in advance in the management terminal 81 and set in the L2 switch 21 when a trouble that causes a problem occurs.

1, 2, 3, 4, 5, 6, 7 Lines 11, 12, 13 User terminal 21, 22, 23 L2 switch 31, 32, 33 L3 switch 41, 42, 43 Link aggregation 51, 52 User communication path 61, 62 Communication path 71, 72 Notification path 81 Network management terminal

Claims (6)

A plurality of L2 switches each connected to one or more user terminals;
A plurality of L3 switches to which the plurality of L2 switches are connected, and a line between each of the plurality of L2 switches and the plurality of L3 switches is bundled by link aggregation, and the plurality of L3 switches are In a network system with an interconnected configuration,
The L3 switch is
When a failure of a line directly connected to itself is detected, and a line connecting the plurality of L3 switches to each other is a virtual line, a monitoring packet destined for a predetermined L3 switch is transmitted, and a response signal Means for detecting failure of the virtual circuit according to the presence or absence of reception;
When all of the redundant lines and virtual lines that transfer frames to a predetermined L2 switch among the plurality of L2 switches are detected as failed, the failure is notified to L2 switches other than the predetermined L2 switch. And means for
The L2 switch is
Due to the notification of the failure, the transfer frame destined for the user terminal connected to the predetermined L2 switch is excluded from the transfer frame distribution process at the time of load balancing of the link aggregation, and is sent to the L3 switch not reporting the failure. A network system comprising means for setting to transmit only from a connected line. The network system according to claim 1 ,
The means for notifying a failure from the L3 switch uses a management protocol to notify a failure state or uses a remote connection protocol without notifying the failure itself, and sends a frame to an L2 switch other than the predetermined L2 switch. A network system characterized in that it is configured to change a setting for switching a line for transferring data. The network system according to claim 1 ,
The means for notifying the failure of the L3 switch is configured to notify an L2 switch other than the predetermined L2 switch via a terminal that manages network information. A plurality of L2 switches each connected to one or more user terminals;
A plurality of L3 switches to which the plurality of L2 switches are connected, and a line between each of the plurality of L2 switches and the plurality of L3 switches is bundled by link aggregation, and the plurality of L3 switches are In a network failure avoiding method for a network system having a mutually connected configuration,
The L3 switch is
When a failure of a line directly connected to itself is detected, and a line connecting the plurality of L3 switches to each other is a virtual line, a monitoring packet destined for a predetermined L3 switch is transmitted, and a response signal Detects the failure of the virtual circuit according to the presence or absence of reception,
When all of the redundant lines and virtual lines that transfer frames to a predetermined L2 switch among the plurality of L2 switches are detected as failed, the failure is notified to L2 switches other than the predetermined L2 switch. To do and
The L2 switch is
Due to the notification of the failure, the transfer frame destined for the user terminal connected to the predetermined L2 switch is excluded from the transfer frame distribution process at the time of load balancing of the link aggregation, and is sent to the L3 switch not reporting the failure. A network failure avoidance method characterized in that transmission is performed only from the connected line. The network failure avoidance method according to claim 4 ,
For notification of failure from the L3 switch, the management protocol is used to notify the failure state, or the remote connection protocol is used without notifying the failure itself, and the frame is transferred to an L2 switch other than the predetermined L2 switch. A network failure avoidance method characterized by changing a setting for switching a line to be used. The network failure avoidance method according to claim 4 ,
The failure notification from the L3 switch is notified to an L2 switch other than the predetermined L2 switch via a terminal that manages network information.

Images