CN101321130A - Method, system and device for forwarding traffic - Google Patents

Abstract

The invention discloses a method, system and device for forwarding traffic, belonging to the technical field of communication. The method includes: receiving broadcast traffic from an ingress port, searching for an egress port corresponding to the traffic; when the TM corresponding to the ingress port is different from the TM corresponding to the egress port, judging whether the egress port is a member port of a bundled port; if so, obtaining the Bind the EFUs corresponding to all the member ports in the port, copy the traffic for each EFU in the EFU, and forward the copied traffic to each EFU. The system includes: forwarding device and EFU. The device includes: a receiving module, a searching module and a replicating module. The invention greatly shortens the flow loss time when the Trunk member port is switched, and makes the Trunk member port switch time meet the requirement of the carrier level.

Description

Method, system and device for forwarding traffic

technical field

The present invention relates to the technical field of communications, in particular to a method, system and device for forwarding traffic.

Background technique

Trunk is a bundling technology that bundles multiple physical ports of a network switching device into one logical port for use. By configuring the logical port, the user can realize various routing protocols and services such as VLAN (Virtual Local Area Network, virtual local area network), MPLS (Multi Protocol Label Switching, multi-protocol label switching) and VPN (Virtual Private Network, virtual private network). The physical ports bound together are called members of the Trunk, and all configurations on the logical ports of the Trunk will eventually be converted into configurations on members. The traffic whose egress is a Trunk logical port usually selects one of the member ports through a hash (HASH) operation to send out. Trunk bundling technology can increase the bandwidth of network switching equipment, so that the traffic load is balanced. When a link in the trunk fails, traffic on the link is automatically switched to other available links in the trunk, thereby improving link reliability.

Ethernet (Ethernet) Trunk is a Layer 2 bundling technology that bundles multiple Ethernet ports into a Trunk logical port. Ethernet Trunk can manually establish a connection or dynamically establish and maintain a connection through LACP (Link Aggregation Control Protocol, Link Aggregation Control Protocol)). In an Ethernet switching device, it generally includes multiple EFU (Express Forwarding Unit, fast forwarding unit) and TM (Traffic Manager, traffic manager). The EFU is connected to the port of the device and performs the forwarding action according to the forwarding table; the TM splits the forwarded uplink and downlink, receives traffic from the EFU, and is responsible for traffic scheduling, replication and forwarding according to the ELB (Egress Link Block, downlink table). When the traffic received by the TM from the EFU is broadcast traffic, according to the ELB table in the TM, multiple copies of the traffic are sent to the EFU where each outgoing port connected to the ELB table is located; when the traffic received by the TM from the EFU has a specified outgoing port port, copy the traffic to the TM corresponding to the EFU where the egress port is located, and send it out from the egress port through the EFU. If the EFU where the specified egress port is located is the EFU that receives the traffic, the TM does not need to copy the received traffic , forwarded directly to the specified egress port through the EFU.

In the prior art, when a member port in the Trunk fails, other member ports in the Trunk are selected to be responsible for traffic forwarding. Referring to FIG. 1, the Ethernet switching device has three ports P1, P2 and P3, where P2 and P3 belong to a Trunk, and P1 and the Trunk belong to the same broadcast domain. The ELB tables of TM1 and TM2 store the outbound port information selected according to the algorithm. For example, according to the hash algorithm for a certain traffic, the outbound port is P3. When P3 fails at a certain time t1, the upper layer software of the device refreshes the Trunk table at time t2 according to the faulty port information, and deletes P3 in the table. Correspondingly, TM1 refreshes its own ELB table at time t3 according to the new trunk table, and changes the outgoing port to P2 according to the hash algorithm. After that, the traffic entering TM1 will be sent to EFU1 where P2 is located, and will no longer be sent to TM2, thus completing port traffic switching.

After analyzing the prior art, the inventor found that: when the corresponding Trunk member ports before and after port traffic switching are not on the same EFU, the network switching device will lose traffic during the switching process. If the Trunk joins more VLANs , it will cause the traffic of the network switching device to be interrupted for a long time. As shown in Figure 1, during the period from t1 to t3 when the trunk member port is switched, the traffic of the device will still be sent to the faulty port P3, which will cause traffic loss.

Contents of the invention

In order to shorten the time of traffic loss when a trunk member port is switched, the embodiments of the present invention provide a method, system and device for forwarding traffic. Described technical scheme is as follows:

A method for forwarding traffic, the method comprising:

Receive broadcast traffic from the ingress port, and find the egress port corresponding to the traffic;

When the traffic manager corresponding to the ingress port is different from the traffic manager corresponding to the egress port, judging whether the egress port is a member port of a bundled port;

If so, obtain EFUs corresponding to all member ports in the bundled port, copy the traffic for each EFU in the EFU, and forward the copied traffic to each EFU.

A system for forwarding traffic, the system comprising:

A forwarding device, configured to receive broadcast traffic from an ingress port, and search for an egress port corresponding to the traffic; when the traffic manager corresponding to the ingress port is different from the traffic manager corresponding to the egress port, determine that the egress port Whether it is a member port of a bundled port; if yes, obtain the fast forwarding unit EFU corresponding to all member ports in the bundled port, copy the traffic for each EFU in the EFU, and forward the copied traffic to said each EFU;

The EFU is configured to receive the traffic sent by the forwarding device.

A device for forwarding traffic, the device comprising:

A receiving module, configured to receive broadcast traffic from an ingress port;

A search module, configured to search for an outbound port corresponding to the traffic received by the receiving module;

Copying module, for when the traffic manager corresponding to the ingress port is different from the traffic manager corresponding to the egress port, judge whether the egress port found by the search module is a member port of a bundled port; if yes, then Obtain fast forwarding units EFUs corresponding to all member ports in the bundled port, copy the traffic for each EFU in the EFU, and forward the copied traffic to each EFU.

In the embodiment of the present invention, by copying multiple copies of traffic, the copied traffic is sent to the EFU corresponding to all member ports in the bundled port, so that when the trunk member port is switched, the EFU connected to the outgoing port can directly forward according to the copied traffic, thereby Traffic loss is avoided. Compared with the EFU connected to the outgoing port in the prior art, which needs to wait for the TM corresponding to the incoming port to refresh its own ELB table before forwarding correctly, it greatly shortens the time for traffic loss when the Trunk member port is switched. The switching time of the Trunk member ports meets the requirements of the carrier level, and compared with the prior art, the traffic of the network switching equipment is more effectively protected.

Description of drawings

FIG. 1 is a schematic diagram of traffic forwarding during Trunk port switching in the prior art;

FIG. 2 is a flowchart of a method for forwarding traffic provided by an embodiment of the present invention;

FIG. 3 is a schematic diagram of replication traffic provided by an embodiment of the present invention;

FIG. 4 is a schematic diagram of replication traffic forwarding after a port fails according to an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of a system for forwarding traffic provided by an embodiment of the present invention;

FIG. 6 is a schematic structural diagram of a device for forwarding traffic provided by an embodiment of the present invention;

Fig. 7 is another schematic structural diagram of the device for forwarding traffic provided by the embodiment of the present invention.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the implementation manner of the present invention will be further described in detail below in conjunction with the accompanying drawings.

An embodiment of the present invention provides a method for forwarding traffic, including: receiving broadcast traffic from an ingress port, searching for an egress port corresponding to the traffic; when the TM corresponding to the ingress port is different from the TM corresponding to the egress port, judging whether the port is a member port of a bundled port; if yes, obtain the EFUs corresponding to all member ports in the bundled port, copy the traffic for each EFU in the EFU, and forward the copied traffic to each EFU .

Referring to Fig. 2, the method for forwarding traffic provided by the embodiment of the present invention may specifically include:

201: The TM in the network switching device receives broadcast traffic from an ingress port, and searches for an egress port corresponding to the traffic.

Specifically, after receiving a certain broadcast traffic, the TM can search its own ELB table to find the outbound port corresponding to the broadcast traffic, for example, the outbound port corresponding to the traffic whose destination address is MAC1 is P3. Among them, the outbound port information in the TM's ELB table is based on the Trunk ID found in the routing table and the hash value obtained by calculating the parameters in the traffic (such as the source IP address) according to the preset hash algorithm , obtained from the selected member ports in the corresponding Trunk table.

202: When the TM corresponding to the egress port is not the above-mentioned TM, the above-mentioned TM judges whether the egress port is a member port of the bundled port, and if the egress port is a member port of the bundled port, execute 203; otherwise, execute 205.

Specifically, the TM can obtain information about whether the outgoing port belongs to a bundled port from the ELB table. Usually, the ELB table stores the identification information of the outgoing port, which is used to identify whether the outgoing port belongs to a bundled port. Whether the port is a member port of the port bundle.

203: The above-mentioned TM obtains the EFUs corresponding to all the member ports in the above-mentioned bundled port.

For example, there are five member ports P2, P3, P4, P5, and P6 in the bundled port, among which, P2 and P3 are connected to EFU2, P4 is connected to EFU3, and P5 and P6 are connected to EFU4, then the TM that receives the traffic obtains There are three EFUs corresponding to the five member ports in the bundled port, namely: EFU2, EFU3, and EFU4.

204: The above-mentioned TM copies the above-mentioned flow for each EFU in the obtained EFU, and forwards the copied flow to each EFU, and then ends.

In the embodiment of the present invention, the traffic sent to the EFU connected to the egress port is called the main traffic, and the traffic sent to the EFUs connected to other member ports in the bundled port is called the protection traffic. For example, referring to FIG. 3 , the network switching device has three ports P1 , P2 and P3 , where P2 and P3 belong to a Trunk, and P1 and the Trunk belong to the same broadcast domain. When TM1 receives some traffic from port P1, it finds that the outbound port corresponding to the traffic is P3 in its ELB table, and knows that P3 belongs to the bundled port, then TM1 is the EFU1 and EFU1 corresponding to the member port P2 in the bundled port. EFU2 corresponding to member port P3 copies a copy of the traffic and forwards the copied traffic. Among them, the traffic forwarded to EFU2 is the main traffic, and the traffic forwarded to EFU1 is the protection traffic.

205: The above TM forwards the traffic through the EFU connected to the egress port, and then ends.

Wherein, the forwarding process can be specifically as follows:

The above TM copies the received traffic and sends it to the TM corresponding to the above outbound port. After the TM corresponding to the outbound port receives it, it sends it to the EFU connected to the outbound port. After the EFU receives the traffic, it sends the traffic to the outbound port. port.

For example, TM1 receives traffic from P1, the outgoing port is P3, and P3 is not a bundled port, P1 is connected to EFU1, and P3 is connected to EFU2, then TM1 copies the traffic and sends it to EFU2 through TM2, and EFU2 forwards it after receiving it to port P3; if the outgoing port is P2, and P2 is not a bundled port, and both P2 and P1 are connected to EFU1, then TM1 does not need to copy the received traffic, but directly forwards the traffic to EFU1, and EFU1 forwards the traffic after receiving it to port P2.

Further, the above method may also include:

After each EFU receives the traffic forwarded by TM, it looks for the outbound port corresponding to the traffic, and judges whether the found outbound port is the port it is connected to, and if so, sends the traffic to the outbound port; otherwise, discards it the traffic. Specifically, the EFU can look up the corresponding Trunk table according to the Trunk ID found in the routing table, and select a corresponding The member port of the port is used as the outgoing port. By judging and forwarding, the EFU receiving the main traffic can forward the traffic to the port, and other EFUs discard the received protection traffic, so as to ensure that the traffic sent by the network switching device is only the main traffic, and the protection traffic will not be sent.

In addition, the above method may also include:

When the outbound port found by the TM fails and the outbound port belongs to a bundled port, the network switching device updates the Trunk table of the bundled port. For example, delete the failed member port in the Trunk table, see Table 1, which is a Trunk table in the network switching device, its length is 8, three physical ports P2, P3 and P4 are bound in this Trunk, and the index number is The hash value obtained by the hash algorithm. When member port P3 fails, delete member port P3 in the Trunk table, and reorder the remaining member ports. There are many ways to sort. In the updated Trunk table shown in Table 2, delete member port P3 Finally, there are only member ports P2 and P4 in the table, and P2 and P4 are arranged in ascending order according to the corresponding index numbers from small to large.

Table 1

The index number member port 0 P2 1 P3 2 P4 3 P2 4 P3 5 P4 6 P2 7 P3

Table 2

The index number member port 0 P2 1 P4 2 P2 3 P4 4 P2 5 P4 6 P2 7 P4

For example, referring to Fig. 4, when P3 fails at a certain time t1, the upper layer software of the device refreshes the Trunk table at time t2 according to the faulty port information, as shown in Table 2. Correspondingly, after receiving the traffic forwarded by TM1, EFU1 and EFU2 calculate the hash value as 1, and search the Trunk table respectively. If the Trunk table is not updated at this time, find that the outbound port corresponding to the traffic is P3. If this When the Trunk table has been updated, it is found that the outbound port corresponding to this traffic is P4, so the replication traffic on EFU3 connected to P4 becomes the main traffic, and the replication traffic on other EFUs becomes protection traffic. For the case where the Trunk table has been updated, that is, the application scenario shown in Figure 4, EFU3 judges that the found outbound port P4 is the port it is connected to, and then sends the traffic to outbound port P4; EFU1 and EFU2 judge that the outbound port Port P4 is not connected to itself, so this traffic is dropped.

Further, after the network switching device updates the Trunk table, the TM can also update the ELB table, that is, the above method also includes:

When the outbound port found by the TM fails and the trunk table has been updated, the TM corresponding to all member ports in the bundled port will select a new outbound port from the updated trunk table according to the same algorithm, and add its own ELB table The outbound port in is updated to the selected new outbound port. As shown in Figure 4, TM1, TM2, and TM3 all obtain the same hash value 1 according to the same hash algorithm, select the corresponding new outbound port P4 in the updated Trunk table, and then record the generated ports in their respective ELB tables. The faulty original outgoing port P3 is updated to the new outgoing port P4. At this time, the process of updating the ELB table does not affect the forwarding of the main traffic, thereby shortening the time for traffic loss when the trunk member port is switched.

In this embodiment, by copying multiple copies of traffic, the copied traffic is sent to the EFUs corresponding to all member ports in the bundled port, so that when the trunk member port is switched, the EFU connected to the outgoing port can directly forward according to the copied traffic, thereby avoiding Compared with the EFU connected to the outgoing port in the prior art, which needs to wait for the TM corresponding to the incoming port to refresh its own ELB table before forwarding correctly, this greatly shortens the time for traffic loss when the Trunk member port is switched, making The switching time of Trunk member ports meets the carrier-class requirements. Compared with the existing technology, it saves the time for TM to refresh the ELB table (the time interval is usually relatively long, and generally the update time for a port in an ELB table by TM is as long as about 100ms), and shortens the time for switching trunk member ports It is the time from the failure of the port to the update of the trunk table. The time interval is very small, generally less than 10ms, and is not affected by the number of VLANs added to the trunk, thus more effectively protecting the traffic of the network switching device. After being judged by the EFU, it is forwarded to ensure that the network switching device only forwards a copy of the traffic (that is, the main traffic) from the egress port, while the other copy traffic (that is, the protection traffic) is not forwarded, which prevents the network switching device from forwarding multiple copies. flow.

Referring to Fig. 5, the embodiment of the present invention also provides a system for forwarding traffic, which specifically includes:

The forwarding device 501 is used to receive broadcast traffic from the ingress port, and search for the egress port corresponding to the traffic; when the TM corresponding to the ingress port is not the same as the TM corresponding to the egress port, determine whether the egress port is a member port of the bundled port; if so, Then obtain the EFUs corresponding to all member ports in the bundled port, copy the traffic for each EFU in the EFU, and forward the copied traffic to each EFU;

The EFU 502 is configured to receive traffic sent by the forwarding device 501.

Further, EFU 502 can also include:

The forwarding module is used to find out the corresponding outgoing port of the traffic received by the EFU 502, and judge whether the port is the port connected to the EFU 502, and if so, send the traffic to the outgoing port; otherwise, discard the traffic.

Further, the forwarding device 501 may also include:

The updating module is used to select a member port other than the outgoing port from the bundled ports as a new outgoing port according to a preset algorithm when the outgoing port fails, and update the outgoing port in the downlink list of the forwarding device 501 to New out port.

The forwarding device 501 in this embodiment may specifically be a TM in a network switching device.

In this embodiment, by forwarding and copying multiple copies of traffic, the copied traffic is sent to the EFUs corresponding to all member ports in the bundled port, so that when the trunk member port is switched, the EFU connected to the outgoing port can directly forward according to the copied traffic, thereby Traffic loss is avoided. Compared with the EFU connected to the outgoing port in the prior art, which needs to wait for the TM corresponding to the incoming port to refresh its own ELB table before forwarding correctly, it greatly shortens the time for traffic loss when the Trunk member port is switched. Make the switchover time of Trunk member ports meet the carrier-class requirements. Compared with the existing technology, it saves the time for TM to refresh the ELB table, shortens the time for trunk member port switching to the time from port failure to trunk table update, and more effectively protects the traffic of network switching devices. In addition, forwarding after judgment by the EFU ensures that the network switching device only forwards a copy of the copied traffic (ie, the main traffic) from the egress port, while other duplicated traffic (ie, the protection traffic) is not forwarded.

Referring to Fig. 6, the embodiment of the present invention also provides a device for forwarding traffic, which specifically includes:

A receiving module 601, configured to receive broadcast traffic from an ingress port;

A search module 602, configured to search for an outbound port corresponding to the traffic received by the receiving module 601;

Copying module 603, is used for when the traffic manager corresponding to the incoming port is different from the traffic manager corresponding to the outgoing port, judging whether the outgoing port found by the search module 602 is a member port of the bundled port; The EFU corresponding to all the member ports copies the flow received by the receiving module 601 for each EFU in the EFU, and forwards the copied flow to each EFU.

Further, referring to FIG. 7, the above-mentioned device also includes:

The updating module 604 is used to select a member port other than the outgoing port from the bundled ports as a new outgoing port according to a preset algorithm when the outgoing port found by the searching module 602 fails, and add it to the ELB table of the device. The outbound port of is updated to the new outbound port.

The device for forwarding traffic in this embodiment may specifically be a TM in a network switching device.

In this embodiment, multiple copies of traffic are copied through TM, and the copied traffic is sent to the EFU corresponding to all member ports in the bundled port, so that when the trunk member port is switched, the EFU connected to the outgoing port can directly forward according to the copied traffic, thereby Traffic loss is avoided. Compared with the EFU connected to the outgoing port in the prior art, which needs to wait for the TM corresponding to the incoming port to refresh its own ELB table before forwarding correctly, it greatly shortens the time for traffic loss when the Trunk member port is switched. Make the switchover time of Trunk member ports meet the carrier-class requirements. Compared with the existing technology, it saves the time for TM to refresh the ELB table, shortens the time for trunk member port switching to the time from port failure to trunk table update, and more effectively protects the traffic of network switching devices.

The embodiment of the present invention can be implemented by combining software and hardware, and the corresponding software program can be stored in a readable storage medium, for example, a hard disk, cache or optical disk of the switch.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the protection of the present invention. within range.

Claims (8)

1. the method for a converting flow is characterized in that, described method comprises:

Receive broadcast traffic from inbound port, search described flow corresponding output port;

When described inbound port corresponding flow manager and described outbound port corresponding flow manager are inequality, judge whether described outbound port is the member port of bond ports;

If, then obtain the quick retransmission unit EFU of member port correspondences all in the described bond ports, for each EFU among the described EFU duplicates described flow, and the flow of transmitting after duplicating is given described each EFU.

2. the method for converting flow according to claim 1 is characterized in that, described method also comprises:

After described each EFU receives described flow, search described flow corresponding output port;

Judge the described outbound port that self finds whether by self the port of company, if then described flow is mail to described outbound port; Otherwise, abandon described flow.

3. the method for converting flow according to claim 1 is characterized in that, described method also comprises:

When described outbound port breaks down, described all member port corresponding flow managers are all chosen member port except that described outbound port as new outbound port according to identical algorithm from described bond ports, and the described outbound port in the downlink table of oneself is updated to described new outbound port.

4. the system of a converting flow is characterized in that, described system comprises:

Retransmission unit is used for receiving broadcast traffic from inbound port, searches described flow corresponding output port; When described inbound port corresponding flow manager and described outbound port corresponding flow manager are inequality, judge whether described outbound port is the member port of bond ports; If, then obtain the quick retransmission unit EFU of member port correspondences all in the described bond ports, for each EFU among the described EFU duplicates described flow, and the flow of transmitting after duplicating is given described each EFU;

EFU is used to receive the described flow that described retransmission unit is sent.

5. the system of converting flow according to claim 4 is characterized in that, described EFU also comprises:

Forwarding module is used to search the described flow corresponding output port that described EFU receives, judge described outbound port whether by described EFU the port of company, if then described flow is mail to described outbound port; Otherwise, abandon described flow.

6. the system of converting flow according to claim 4 is characterized in that, described retransmission unit also comprises:

Update module, be used for when described outbound port breaks down, from described bond ports, choose the new outbound port of a member port conduct except that described outbound port according to default algorithm, and the described outbound port in the downlink table of described retransmission unit is updated to described new outbound port.

7. the device of a converting flow is characterized in that, described device comprises:

Receiver module is used for receiving broadcast traffic from inbound port;

Search module, be used to search the flow corresponding output port that described receiver module is received;

Replication module is used for when described inbound port corresponding flow manager and described outbound port corresponding flow manager are inequality, judges describedly to search whether the outbound port that module finds is the member port of bond ports; If, then obtain the quick retransmission unit EFU of member port correspondences all in the described bond ports, for each EFU among the described EFU duplicates described flow, and the flow of transmitting after duplicating is given described each EFU.

8. the device of converting flow according to claim 7 is characterized in that, described device also comprises:

Update module, be used for when described when searching outbound port that module finds and breaking down, from described bond ports, choose the new outbound port of a member port conduct except that described outbound port according to default algorithm, and the described outbound port in the downlink table of described device is updated to described new outbound port.

Images