KR20120075013A - Method for avoiding unicast flooding in a redundant network therfor network system - Google Patents

Abstract

PURPOSE: A unicast floating prevention method in a duplex network and network system thereof are provided to automatically and rapidly unicast floating generated from link errors or restoration process by using a searching procedure supported by a duplex protocol. CONSTITUTION: A backbone switch unit searches a link between one or more adjacent backbone switch units and an inter-layer switch unit(312). When link errors are detected, the backbone switch unit increases count values as m or 1(314,316). The count value m reaches to the established maximum values, the backbone switch unit clears an address table(318,326). The backbone switch unit executes a update procedure in order to register the MAC(Medium Access Control) address of network apparatuses in the address table(328).

Description

METHOD FOR AVOIDING UNICAST FLOODING IN A REDUNDANT NETWORK THERFOR NETWORK SYSTEM}

The present invention relates to a method for preventing unicast floating in a redundant network and a network system for the same, and more particularly, to a switch control method for removing unicast floating and a network system for the same.

In the early days of the Internet, networks that provided 10 Mbps unidirectional speed services ran at terabit speeds via Fast Ethernet, Gigabit Ethernet, and 10 Gigabit Ethernet. It is evolving into a network that can support even. The driving force behind this development is the human desire for better content and faster speeds.

Due to these human desires, the network is realizing “ALL IP” to deliver all information such as data, voice, video, etc. in Internet Protocol packets. For example, the data includes personal or financial information such as Internet banking in addition to general Internet information, and voice can be used as an internet phone service that can be used instead of a regular phone, and video can be provided as an IPTV service in addition to general VOD.

The reality is that IP networks are becoming an important infrastructure that carries almost all information. Therefore, it is clear that disconnection or failure of the IP network can lead to severe service interruption, which can cause huge financial loss for the service provider.

As a result, companies are designing and building networks to minimize service disruptions, and endeavor to operate them. A representative example is the network design and operation of redundancy of data centers and network devices in data centers.

Since a data center is typically a place where all information of the enterprise is not only concentrated and switched, but also important service systems are gathered, the failure of the data center means the interruption of the entire enterprise. As a result, companies are minimizing service disruptions by geographically separating and redundancy of data centers.

In addition, all network equipment such as routers, L2 / L3 switches, L4 switches, firewalls, and Intrusion Prevention System (IPS) in the data center are redundantly configured to minimize service interruption due to equipment failure.

Finally, a protocol is used to duplicate the default route information of the personal computer for redundant equipment.

Typical protocols for this include the Virtual Router Redundancy Protocol (VRRP), the Hot Standby Router Protocol (HSRP), and the Gateway Load Balancing Protocol (GLBP). This protocol is a network device protocol that allows service to be performed through the remaining network device automatically even when a problem occurs in one network device among the redundant network devices.

However, when there is only one network device, it is not necessary to consider it, but when the network device is redundant, the asymmetry of the traffic path must be considered. For example, in a redundant network, when the outgoing traffic path and the incoming traffic path do not coincide with each other, when the traffic comes in from another network device, the packet flows even though it is not broadcast but unicast. Flooding).

This packet floating is a cause of occupying the bandwidth of another user or server, which can cause service interruption in severe cases and must be eliminated. In the case of a corporate network with a small amount of such a floating packet, it can be operated as it is, but a corporate network with a large amount of floating packet causes a problem such as poor server performance.

Therefore, companies are preparing a plan to prevent the performance of the server due to the floating packet. For example, by setting the MAC address-out time (MAC Aging-Out Time) of the second layer (Layer 2) to the same as the address table update time (ARP Time-Out Time) of the third layer (Layer 3), It is preparing for the case where there is no topology change due to link failure.

The MAC Aging-Out Time is a time for maintaining a MAC Address of a network device stored in an address table. For example, the MAC address for which packet transmission is not performed until the MAC Aging-Out Time elapses is deleted from the address table.

The ARP Time-Out Time is a time for updating the address table periodically or aperiodically. For example, whenever the ARP time-out time elapses, the address table managing the MAC address is updated.

Floating of the original packet refers to a phenomenon in which a unicast packet is sent to all ports except the received port when the broadcast packet is a broadcast packet and when the destination MAC address is not registered in the switch.

In this case, each server and the terminal are adversely affected by the degradation of service performance by receiving packets that they do not need to receive. This may vary depending on the traffic volume, but basically degrades service performance and, in severe cases, TCP connections may be lost.

In general, however, a function to prevent unicast flooding may be included in some devices, but changes in network topology are not considered. That is, unicast floating that occurs when the topology changes due to link failure or recovery of the switch network cannot be prevented and additional measures are required.

In general, link failure in a network can be caused by not only physical disconnection of the optical cable, but also failure of the module of the network equipment or failure of the Gigabit Interface Converter (GBIC) component that converts the module's Ethernet information into an optical signal. It can also be caused by operator error or software problems with the equipment.

Therefore, there is an urgent need to prepare a plan to prevent unicast floating in the topology change due to link failure or recovery as well as the case where there is no topology change due to a link failure in a switch network in a redundant network.

A preferred embodiment according to the present invention proposes a method for preventing unicast floating caused by a link failure in a redundant network and a network system for the same.

In addition, a preferred embodiment according to the present invention proposes a method for updating an address table when a link failure connecting adjacent backbone switches occurs in a redundant network and re-updates the address table when the failed link is restored, and a network system for the same. do.

In addition, a preferred embodiment according to the present invention proposes a method for setting the registered MAC address holding time and the address table update time to be the same, and updating the address table when a failure or failure recovery between adjacent backbone switches occurs, and a network system therefor. .

In addition, a preferred embodiment according to the present invention proposes a method and network system for maintaining a user service performance by quickly and automatically eliminates unicast floating occurs in the event of a failure by using ICMP search without operator involvement do.

According to a preferred embodiment of the present invention, a method for preventing unicast floating in a backbone switch of a redundant network may be performed by continuously specifying a predetermined maximum number of times or more based on a registered MAC address holding time and an address table updating time set to the same time. When a search for a link with an adjacent backbone switch or a specific interlayer switch fails, a link failure processing process of updating an internal address resolution protocol table or a neighbor search protocol table, and the address resolution protocol table or the neighbor search protocol After updating the table, successively succeeding more than a preset maximum number of times for searching for a link with the specific neighboring backbone switch or the specific interlayer switch based on the registered MAC address holding time and the address table updating time set to the same time. , Includes a link recovery process for re-updating the address resolution protocol table or the neighbor discovery protocol tables.

In addition, according to a preferred embodiment of the present invention, a redundant network system for preventing unicast floating includes at least one neighboring backbone switch, at least one interlayer switch, and a registered MAC address holding time and address table set to the same time. A backbone switch for discovering a link with the at least one adjacent backbone switch or the at least one interlayer switch based on an update time,

The backbone switch,

Determining an existing address when a search for a link with a specific neighboring backbone switch or a specific interlayer switch fails more than a preset maximum number of times consecutively based on the registered MAC address holding time and the address table update time set to the same time Perform a link failure handling procedure that updates the protocol table or the neighbor discovery protocol table,

After updating the address resolution protocol table or the neighbor search protocol table, searching for a link with the specific neighbor backbone switch or the specific interlayer switch based on the registered MAC address holding time and the address table update time set to the same time; If a successive succession more than a predetermined number of times in succession, characterized in that for performing the link recovery processing to re-update the address resolution protocol table or the neighbor search protocol table.

In the preferred embodiment of the present invention, by using the discovery procedure supported by the duplication protocol, it is possible to quickly and automatically remove unicast floating occurring at the time of link failure or recovery without operator's intention, thereby maintaining user service performance. This can significantly reduce network management costs.

1 shows a general example of a redundant network;
2 illustrates an example of a redundant network that supports ICMP probing to implement a preferred embodiment of the present invention;
3 illustrates a control flow performed in a backbone switch according to an exemplary embodiment of the present invention;
4 is a diagram illustrating a processing procedure when a failure occurs in a redundant network according to a preferred embodiment of the present invention;
5 is a diagram illustrating a processing procedure at the time of failure recovery in a redundant network according to a preferred embodiment of the present invention.

In the following description of the preferred embodiment, when it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the technical gist, the detailed description thereof will be omitted.

In the preferred embodiment of the present invention to be described later will provide a way to flexibly cope with the change in the network topology unicast floating by asymmetric routing caused in a typical corporate network. In the preferred embodiment of the present invention for this purpose, the ICMP search function is used. The ICMP search function refers to a function for searching for a network device to which adjacent network devices, that is, network devices directly connected by a link.

In addition, the preferred embodiment of the present invention can be applied to various networks. For example, it can be applied to both IPv4 and IPv6 networks.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 shows a general example of a redundant network.

Referring to FIG. 1, an enterprise network includes a backbone area, a switch area, and a server area. In this actual corporate network configuration, the segment with users or servers belongs to a specific VLAN, and many devices are connected to each VLAN. The user area or the server farm area forms a sufficient redundancy by connecting the backbone area 110 and the redundant network equipment in a mesh form.

In addition, since most of the routing protocols such as the backbone region 110 and OSPF are used, the traffic from the backbone position to the server farm regions 120 and 130 is load-balanced by hash calculation. In addition, the server communicates with the external network by sending traffic to one of the redundant L3 switches using VRRP, which is a gateway redundancy protocol.

The floating in such a corporate network is caused by asymmetric routing in the redundant configuration by VRRP.

For example, referring to FIG. 1, it is assumed that the server A passes through the switch A when going up to the backbone area 110, and the case goes down through the switch B when going down from the backbone area 110 to the server A.

In the above assumption, the switch B does not know the MAC address of the server A if the situation in which there is no packet to transmit to the server A is maintained for the MAC MAC-Out Time (usually 5 minutes). Therefore, after that, the packet to be transmitted to the server A is floated. At this time, the server B in the same VLAN is also delivered to the server A unnecessarily, affecting the normal traffic processing of the server B, thereby reducing the service performance of the server B.

2 shows an example of a redundant network that supports ICMP discovery to implement a preferred embodiment of the present invention.

Referring to FIG. 2, two adjacent backbone switches 210, 220 are linked through a trunk 230. In this case, the adjacent backbone switches 210 and 220 mean a backbone switch in which a direct link is formed through a trunk without any other configuration in the middle. The adjacent backbone switches 210 and 220 confirm that they exist adjacent to each other through a periodic or aperiodic ICMP search.

Meanwhile, the backbone switches 210 and 220 have a link formed through a trunk with at least one interlayer switch among a plurality of interlayer switches existing for each floor. For example, the backbone switch 210 may include an interlayer switch corresponding to the reference number 242 among the interlayer switches 240 present on the first floor and an interlayer switch and the trunk corresponding to the reference number 252 among the interlayer switches 250 present on the second floor. Linked through. In addition, the backbone switch 220 may include an interlayer switch corresponding to the reference number 244 among the interlayer switches 240 existing on the first floor and an interlayer switch and the trunk corresponding to the reference number 254 among the interlayer switches 250 present on the second floor. Linked through.

In addition, the backbone switches 210 and 220 perform a search procedure periodically or aperiodically for a link with a plurality of interlayer switches existing for each floor. The search procedure monitors whether the corresponding links are normal.

2 shows a path for the second back switch 220 to search for the first backbone switch 210 and the interlayer switches 242 and 244 located on the first floor. At this time, the second backbone switch 220 is ICMP to each device IP address of the VLAN designated for monitoring the first backbone switch 210, the first floor interlayer switches 242 and 244, and the second floor interlayer switches 252 and 254. Perform a search.

Thus, by performing the ICMP search, the second backbone switch 220 not only monitors the link connecting the first and second backbone switches but also monitors the link between the backbone and the interlayer switch.

At this time, if the monitor VLAN is included in the link between the interlayer switches 242 and 244, even if the link between the backbone and the interlayer switch fails, the switch port that was blocked by the Spanning-Tree Protocol is forwarded, and ICMP discovery is performed by the interlayer switch. You will succeed. Therefore, the link failure cannot be detected, so the VLAN for monitoring for ICMP discovery should be removed between the two interlayer switches.

Each of the interlayer switches 242, 244, 252, and 254 is connected to a user terminal or a server as an end device.

Meanwhile, trunks shown in FIG. 2 form a link through a procedure in which each device checks a mutual relationship. For example, forming the link means a state in which mutual devices that form the link acquire and know the address of the other party.

Therefore, a procedure for this is required to form a link between equipments. For example, the Address Resolution Protocol (ARP) in IPv4 and the Neighbor Discovery Protocol (NDP) in IPv6 may be representative.

The ARP is a protocol used to map an IP address to a physical network address on an IP network. Here, the physical network address means a network card address of Ethernet or Token Ring.

For example, if IP host A does not know the physical network address of IP host B when it wants to send an IP packet to IP host B, it uses the ARP protocol to network the ARP packet with the destination IP address B and the broadcast physical network address. Send it on. When IP host B receives an ARP packet whose IP address is at its destination, it responds to A with its physical network address.

The IP addresses collected in this way and the corresponding physical network address information are stored in a table in memory called ARP cache of each IP host. Then, at the next packet transmission, the collected IP addresses and corresponding physical network address information stored in the table are used again.

Unlike the ARP, when the IP host knows its physical network address but does not know the IP address, there is a case where RARP is used to request an IP address from a server.

The NDP is a core protocol of IPv6 and is a protocol for searching for neighboring neighbor nodes. At this time, as an NDP signaling message, a neighbor solicitation (NS) message and a neighbor advertisement message (NA) exist. In addition, the Redirect Message includes a Router Solicitation (RS) message and a Router Advertisement (RA) message. The RS message is an ICMP query message for router discovery. At this time, the router responds with an RA message. The RA message is an announcement message of the ICMP and is a message packet indicating that the router is a router.

The function of the NDP includes address automatic configuration and movement detection as well as ARP function of IPv4. As described above, ARP is an address resolution protocol that converts a logical IP address (network layer) into a physical MAC address (data link layer).

Hereinafter, in the preferred embodiment of the present invention, "ARP" and "NDP" are collectively used as the term "search protocol." In addition, the term 'ARP table' generated by the ARP and 'NDP table' generated by the NDP are collectively used as the term 'address table'.

In addition, the term 'ARP table timeout time' for updating the ARP table in IPv4 and 'NDP table timeout time' for updating the NDP table in IPv6 are collectively referred to as 'address table update time'. use.

In addition, the ARP Request Message sent by the backbone switch to form the link in ARP and the Neighbor Solicitation message sent by the backbone switch to form the link in NDP are collectively called the request message. Use it. In addition, the ARP Response Message transmitted by the interlayer switch for link formation in ARP and the Neighbor Advertisement message transmitted by the interlayer switch for link formation in NDP are called response messages. Commonly used.

3 shows a control flow performed in a backbone switch according to an exemplary embodiment of the present invention.

The control operation shown in FIG. 3 sets the registered MAC address-out time equal to the address table update time (ARP timeout time) so that unicast floating occurs while there is no change in the network. prevent. However, if a network failure occurs in which the link between the network equipment is broken, unicast floating occurs again.

Therefore, in order to prevent unicast floating due to link disconnection between network equipments, the inter-switch ICMP packet is searched to monitor the link between networks. When detecting the link disconnection through the monitoring, the address table of the network device is cleared. Then, by restarting the timer between the third layer (Layer 3) engine and the second layer (Layer 2) engine, the MAC address is re-registered in the address table, thereby eliminating the floating immediately upon detecting the failure.

More specifically, referring to FIG. 3, in step 310, the backbone switch has the same time as the registered MAC address holding time (MAC Aging-Out Time, T1) and the address table update time (ARP timeout time, or NDP timeout time, T2). Set to. A count value m for counting the number of consecutive failure detections for the same link and a count value n for counting the number of recovery detections of a previously failed link are initialized. A detailed description of the conditions for counting the count values m and n will be described later.

On the other hand, setting the T1 and T2 to the same time may be performed only at the time of initial operation (including during initial network construction) or when the corresponding time is changed. The count value m may be initialized during the initial operation (including the initial network construction) or when the number of link failure detections is recovered before reaching the set maximum value count_max1 or when the set maximum value is reached. In addition, the count value n may be initialized during an initial operation (including the initial network establishment) or when the link failure is detected again before the number of link recovery detections reaches the set maximum value (count_max2) or when the set maximum value is reached. .

The maximum value count_max1 set corresponding to m and the maximum value count_max2 set corresponding to n may be different or the same. For example, it is preferable to set the maximum value (count_max1) corresponding to m and the maximum value (count_max2) corresponding to n to a number greater than three.

The backbone switch performs a search procedure for a link with at least one adjacent backbone switch and an interlayer switch in step 312. The search procedure is performed for the purpose of checking whether the link between the backbone switch and the at least one backbone switch or the link between the interlayer switches is normal. For example, the discovery procedure may be performed by the backbone switch sending an ICMP request message to an adjacent backbone switch or an interlayer switch, and receiving an ICMP corresponding message responsive to the ICMP request message from the adjacent backbone switch or an interlayer switch. .

In addition, the backbone switch performs a discovery procedure on the links connecting all the interlayer switches, and checks whether the corresponding links operate normally.

When the discovery procedure is completed as described above, the backbone switch performs a procedure (steps 314 to 318, 326, and 328) related to a link failure and a procedure (steps 320 to 324, 326, and 328) according to link restoration. ).

In FIG. 3, the procedure according to the link failure and the procedure according to the link recovery are performed simultaneously, but the preferred embodiment of the present invention does not limit the order in which the two procedures are performed.

First, referring to a link failure procedure, the backbone switch determines whether there is a failed link among the links with the neighboring backbone switch and the interlayer switch based on the result of the discovery procedure performed in step 314. In this case, the link failure refers to a case where the connection between the two network devices is lost due to a physical cable failure, a GBIC failure, or a module failure of the equipment.

If a link failure is detected, the backbone switch increases the count value m by 1 in step 316 and then determines whether the count value m reaches a preset maximum value count_max1 in step 318. When the count value m reaches a preset maximum value count_max1, it means that a predetermined number of failures are continuously detected on a specific link. For example, it means that an ICMP search for a specific link has failed a predetermined number of times in succession.

In this case, the backbone switch clears the address table in step 326 to delete the previously managed MAC address. In operation 328, the backbone switch performs an update procedure for registering MAC addresses of network devices in an address table. For example, the update procedure may be performed by outputting an ARP request message or NS message through all ports existing in the backbone switch, and receiving an ARP response message or NA message in response. In this case, the transmission of the ARP request message or the NS message may be performed in response to the inflow of traffic.

By updating the address table as soon as a failure is detected as described above, packet transmission can be continued by the newly updated address table, thereby preventing the occurrence of unicast floating due to link recovery. That is, unicast floating or traffic loss caused by network topology changes can be eliminated.

However, if the link failure is not detected in step 314 or failure detection for a specific link is not detected continuously in a predetermined number of times, the backbone switch returns to step 312 to perform a periodic or aperiodic search procedure.

Next, referring to the link restoration procedure, the backbone switch determines whether the link in which the failure occurred during the link with the adjacent backbone switch or the interlayer switch is restored, based on the result of the discovery procedure performed in step 320.

If it is detected that the failed link is restored, the backbone switch increases the count value n by 1 in step 322 and then determines whether the count value n reaches a preset maximum value count_max2 in step 324. When the count value n reaches the preset maximum value count_max2, it means that the recovery for the specific link in which the failure has occurred has been detected continuously for the set number of times. For example, it means that ICMP search for a specific link has succeeded a predetermined number of times in succession.

In this case, the backbone switch deletes the MAC address previously managed by clearing the address table updated due to the failure in step 326. In operation 328, the backbone switch performs a renewal procedure for registering MAC addresses of network devices in an address table. For example, the renewal procedure may be performed by outputting an ARP request message or NS message through all ports present in the backbone switch, and receiving an ARP response message or NA message in response. In this case, the transmission of the ARP request message or the NS message may be performed in response to the inflow of traffic.

By updating the address table as soon as the failure is repaired, packet transmission can be continued by the newly updated address table, thereby preventing the occurrence of unicast floating due to link recovery. That is, unicast floating or traffic loss caused by network topology changes can be eliminated.

However, if the recovery of the failed link is not detected in step 320 or the failure recovery of a specific link is not detected continuously in a predetermined number of times, the backbone switch returns to step 312 to search periodically or aperiodically. Perform

4 shows a processing procedure when a failure occurs in a redundant network according to a preferred embodiment of the present invention.

4 (a) shows an example of an asymmetric path state in a normal situation where no failure occurs. In other words, the uplink traffic transmitted by the first terminal located at the end of the first layer is transmitted through the link connecting the first interlayer switch and the first backbone switch. Downlink traffic input to the second backbone switch is delivered to the first terminal through a path connecting the first backbone switch and the first interlayer switch.

At this time, a discovery procedure is performed between the second backbone switch and the first backbone switch. The discovery procedure monitors whether a failure occurs in a link connecting the second backbone switch and the first backbone switch. That is, the second backbone switch transmits an ICMP request message to the first backbone switch, and the first backbone switch transmits an ICMP reply message to the second backbone switch in response to the ICMP request message.

4 (b) shows an example of a situation where a failure occurs in a link connecting backbone switches by a discovery procedure. That is, when the second backbone switch fails to search for the first backbone switch, which is an adjacent backbone switch, by a predetermined number of consecutive search procedures, the second backbone switch clears an internal address table. When downlink traffic occurs, a procedure for updating the cleared address table is performed. For example, the second backbone switch transmits a request message through all of its output ports and receives a response message corresponding to the request message. The address table is updated by the MAC addresses of the network devices that have sent the response message.

4 (c) shows an example of an asymmetric path state after updating an address table due to a failure. In other words, the uplink traffic transmitted by the first terminal located at the end of the first layer is transmitted through the link connecting the first interlayer switch and the first backbone switch. The downlink traffic input to the second backbone switch is delivered to the first terminal through a path connecting the second interlayer switch located on the first floor.

This eliminates the unicast floating that can occur with interlayer switches located on the second floor due to a failure in the link between the backbone switches.

5 is a flowchart illustrating a processing procedure in a failure recovery in a redundant network according to an exemplary embodiment of the present invention.

5 (a) shows an example of an asymmetric path state after updating an address table due to a failure. In other words, the uplink traffic transmitted by the first terminal located at the end of the first layer is transmitted through the link connecting the first interlayer switch and the first backbone switch. The downlink traffic input to the second backbone switch is delivered to the first terminal through a path connecting the second interlayer switch located on the first floor.

FIG. 5B shows an example when a failed link is restored by a discovery procedure after a failure occurs. That is, a discovery procedure is performed between the second backbone switch and the first backbone switch, and the discovery procedure monitors whether a failure occurred in a link connecting the second backbone switch and the first backbone switch. For example, the second backbone switch sends an ICMP request message to the first backbone switch, and the first backbone switch sends an ICMP reply message to the second backbone switch in response to the ICMP request message.

As described above, when the second backbone switch succeeds in searching for the first backbone switch that is the neighboring backbone switch by a predetermined number of times (for example, three times) in a continuous search procedure, the second backbone switch clears an internal address table. When downlink traffic occurs, a procedure for updating the cleared address table is performed. For example, the second backbone switch transmits a request message through all of its output ports and receives a response message corresponding to the request message. The address table is updated by the MAC addresses of the network devices which have sent the response message, in particular, the path which was not recognized due to the failure of the link.

5 (c) shows an example of an asymmetric path state in a normal situation after the failure is recovered. In other words, the uplink traffic transmitted by the first terminal located at the end of the first layer is transmitted through the link connecting the first interlayer switch and the first backbone switch. Downlink traffic input to the second backbone switch is delivered to the first terminal through a path connecting the first backbone switch and the first interlayer switch.

This eliminates the unicast floating that can occur in the interlayer switch located on the second floor by recovering the fault that occurred in the link between the backbone switches.

While the present invention has been described in connection with what is presently considered to be the most practical and preferred embodiment, it is to be understood that the invention is not limited to the disclosed embodiments, but is capable of various modifications within the scope of the invention. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be defined not only by the appended claims, but also by the equivalents of the claims.

Claims (8)

A method for preventing unicast floating in a backbone switch of a redundant network,
An address resolution protocol that exists internally when a search for a link with a specific neighboring backbone switch or a specific interlayer switch fails more than a preset maximum number of times based on a registered MAC address holding time and an address table updating time set to the same time. Link failure processing to update a table or neighbor search protocol table;
After updating the address resolution protocol table or the neighbor search protocol table, searching for a link with the specific neighbor backbone switch or the specific interlayer switch based on the registered MAC address holding time and the address table update time set to the same time; And link recovery processing of renewing the address resolution protocol table or the neighbor discovery protocol table when successively succeeding more than a preset maximum number of times. The method of claim 1, wherein the link failure process is performed.
Setting the registered MAC address holding time and the address table update time equally;
Performing a search for a link between at least one adjacent backbone switch and at least one interlayer switch based on the same registered MAC address holding time and address table update time;
Clearing the address resolution protocol table or the neighbor search protocol table if there is an adjacent backbone switch or an interlayer switch that has failed to be searched for more than a predetermined number of times in advance by the search;
Resetting a timer between switches in response to the clearing of the address resolution protocol table or the neighbor discovery protocol table, and performing an address re-registration procedure to update the address resolution protocol table or the neighbor discovery protocol table. How to prevent floating. The method of claim 2, wherein the link recovery process comprises:
Performing a search for a link between at least one adjacent backbone switch and at least one interlayer switch based on the same registered MAC address holding time and address table update time;
Clearing the address resolution protocol table or the neighbor search protocol table if the search for the neighboring backbone switch or the interlayer switch that previously failed to be searched succeeds more than a preset maximum number of times;
Resetting the inter-switch timer in response to the clearing of the address resolution protocol table or the neighbor discovery protocol table, and performing an address re-registration procedure to re-update the address resolution protocol table or the neighbor discovery protocol table. How to prevent cast flooding. 4. The method according to any one of claims 1 to 3,
The address table update time is an address resolution protocol timeout time when the redundant network is an IPv4 network, and the address table update time is a neighbor discovery protocol table timeout time when the redundant network is an IPv6 network. How to prevent floating. In a redundant network system for preventing unicast floating,
At least one adjacent backbone switch,
At least one interlayer switch,
A backbone switch searching for a link between the at least one neighboring backbone switch and the at least one interlayer switch based on a registered MAC address holding time and an address table update time set to the same time;
The backbone switch,
Determining an existing address when a search for a link with a specific neighboring backbone switch or a specific interlayer switch fails more than a preset maximum number of times consecutively based on the registered MAC address holding time and the address table update time set to the same time Perform a link failure handling procedure that updates the protocol table or the neighbor discovery protocol table,
After updating the address resolution protocol table or the neighbor search protocol table, searching for a link with the specific neighbor backbone switch or the specific interlayer switch based on the registered MAC address holding time and the address table update time set to the same time; And performing a link recovery processing procedure for re-updating the address resolution protocol table or the neighbor discovery protocol table when successively succeeding more than a preset maximum number of times. The method of claim 5, wherein the backbone switch,
Setting the registration MAC address holding time and the address table update time equally;
Search for a link with at least one adjacent backbone switch and at least one interlayer switch based on the same registered MAC address holding time and address table update time,
If there is a neighboring backbone switch or an interlayer switch that has failed to search continuously for a predetermined maximum number of times by the search, the address resolution protocol table or the neighbor search protocol table is cleared.
The inter-switch timer is reset in response to the clearing of the address resolution protocol table or the neighbor discovery protocol table, and an address re-registration procedure is performed to update the address resolution protocol table or the neighbor discovery protocol table to perform the link failure processing procedure. A redundant network system that performs. The method of claim 6, wherein the backbone switch,
Search for a link with at least one adjacent backbone switch and at least one interlayer switch based on the same registered MAC address holding time and address table update time,
If the search for the adjacent backbone switch or the interlayer switch that previously failed to be searched succeeds more than a preset maximum number of times, the address resolution protocol table or the neighbor search protocol table is cleared.
In response to the clearing of the address resolution protocol table or the neighbor discovery protocol table, a link recovery processing procedure of resetting the inter-switch timer and performing an address re-registration to re-update the address resolution protocol table or the neighbor discovery protocol table is performed. Perform redundant network system 8. The method according to any one of claims 5 to 7,
The address table update time is an address resolution protocol timeout time when the redundant network is an IPv4 network, and the address table update time is a neighbor discovery protocol table timeout time when the redundant network is an IPv6 network. .

Images