WO2012103704A1

WO2012103704A1 - Multicast duplication method, device and system

Info

Publication number: WO2012103704A1
Application number: PCT/CN2011/076263
Authority: WO
Inventors: 王亮; 张健
Original assignee: 华为技术有限公司
Priority date: 2011-06-24
Filing date: 2011-06-24
Publication date: 2012-08-09
Also published as: CN103222230A; CN103222230B

Abstract

The embodiments of the invention provide a multicast duplication method, device and system. The method includes: the downlink module of an egress Traffic Manager (TM) receives, with a circle port, a multicast message sent from a switching network, and sends it to the uplink module; the uplink module of the egress TM searches for the corresponding multicast member ports according to the group identifier carried in the multicast message; the uplink module duplicates the multicast message according to the number of the multicast member ports included in the egress TM, adds the duplicated multicast messages to the unicast message buffering queue according to the priority, and sends them to the downlink module through a local forwarding channel after configuring the scheduler according to the Quality of Service (QoS) of the multicast member ports, to make the duplicated multicast messages sent out. The multicast duplication method, device and system in the embodiments of the invention can realize providing QoS guarantee in the situation of effectively saving the switching network bandwidth, and can further improve the bandwidth utilization rate of the switching network when the switching network performs multicast message duplication for different egress TMs.

Description

Multicast copying method, device and system

Technical field

Embodiments of the present invention relate to the field of communications technologies, and in particular, to a multicast replication method, apparatus, and system. Background technique

Multicast is a traffic model that copies an upstream packet into multiple copies and distributes it to multiple downstream ports. With the promotion of network applications such as HDTV, multicast technology has become more and more widely used, and multicast performance has become one of the important performance indicators of large-scale network devices such as routers and switches. At present, the architecture of a large-scale forwarding system usually consists of multiple Line Cards and a large-capacity switching network. The Packet Processing and Forwarding Engine (PFE) on the line card is responsible for packet parsing and forwarding, and the switching network performs switching. Features. Due to the complexity of the function, the PFE is usually decomposed into two modules: the packet processor (PP) and the traffic manager (TM), where the PP is responsible for packet processing and determining forwarding rules, and the TM is responsible for forwarding, replication, and quality of service (QoS). ) Guaranteed and other functions. The complete process of multicast packets usually includes: multicast ingress port->uplinkΡΡ->ϋ# ΤΜ->switched network->downlinkΤΜ->downstreamΡΡ->multicast outgoing port, where uplink refers to packet The data is transmitted from the data interface to the switching network. The downlink refers to the transmission of packets from the switching network to the data interface. The performance of multicast replication depends mainly on the implementation and performance of the switching network.

In the prior art, the following three methods are generally used for multicast replication. 1 is a schematic flowchart of a first multicast replication method in the prior art, FIG. 2 is a schematic flowchart of a second multicast replication method in the prior art, and FIG. 3 is a third multicast in the prior art. Schematic diagram of the replication method. As shown in FIG. 1 , a multicast replication method in the prior art is uplink replication, that is, all replication work is completed in the uplink module of the multicast source, and multiple multicast packets are copied. The text enters the uplink queue separately and is forwarded through the switching network. When multicast replication is performed in this way, the multicast module for the multicast source is configured to replicate multicast packets for multiple egress ports. This allows a multicast packet to correspond to an egress port. The packet can enter the unicast queue according to the priority carried. The QoS guarantee can be provided by scheduling the multicast packets in the uplink queue according to the congestion status of the egress port and the priority of the packets. However, if multiple egress ports are not on the same line card, the bandwidth consumption of the TM to the switching network is severe. As a bottleneck, for example, 10G multicast source traffic, when there are 4 multicast members on other line cards, will generate 40G switching network line bandwidth requirements. As shown in FIG. 2, another multicast replication method in the prior art is a switching network replication solution, where: the switching network first copies the multicast packet received from the input side line card to the downlink of the multicast member. The TM module further copies the multicast packets forwarded by the switching network to all member ports by the downlink TM module. When the multicast replication is performed in this way, the multicast packets are not formed in the uplink TM module and the switching network where the multicast source resides. The broadcast packet is scheduled, and the multicast packet is forwarded by the multicast member to the multicast network. The multicast packet is copied to all member ports. Cache and scheduling. At the same time, because the upstream multicast source sends packets to the switching network, it is not subject to the scheduling control of a certain downlink port. Therefore, the outbound port can not schedule the multicast packets in the uplink queue. Provide QoS guarantees. As shown in FIG. 3, another multicast replication method in the prior art is a cyclic tree replication scheme, that is, the uplink TM module in which the multicast source is located uses a binary tree-like manner to copy the packet and send it to the switch network. The downlink module of the intermediate node, wherein the downlink module as the intermediate node is not the downlink TM module where the member port of the multicast packet is located, and the downlink TM module as the intermediate node performs the secondary binary tree replication of the packet. If the number of multicast packets is the same as the number of member ports, the number of the multicast packets is the same as that of the member ports. Then, the copied packet is sent to the next intermediate node through the switching network for replication until it is copied to all member ports through a series of loops. When using this method for multicast replication, if there are many member ports, you need to use multiple times. The intermediate node is selected as the replication agent, and one traffic passes through the switching network multiple times, which seriously consumes the bandwidth of the switching network. It can be seen that the prior art multicast replication method cannot provide QoS guarantee while effectively saving the bandwidth of the switching network. Summary of the invention

The purpose of the embodiments of the present invention is to provide a multicast replication method, apparatus, and system for providing QoS guarantee under the condition that the bandwidth of the switching network is effectively saved.

The embodiment of the invention provides a multicast replication method, including:

The downlink module of the egress traffic manager receives the multicast packet sent by the switching network through the recurring port, and sends the multicast packet to the uplink module, where the egress traffic manager includes the uplink module and the downlink module. And a physical port for communicating with the outside, the downlink module is provided with the cyclic port;

The egress module of the egress traffic manager searches for the multicast member port corresponding to the multicast packet in the first uplink multicast replication table according to the group identifier carried in the multicast packet;

The uplink module of the egress traffic manager calculates the number of multicast member ports of the multicast member port corresponding to the egress packet, and according to the egress traffic manager The number of multicast member ports included in the physical port is copied to the multicast packet, so that the number of the copied multicast packets and the multicast member included in the physical port of the egress traffic manager are The number of ports is the same;

The uplink module of the egress traffic manager adds the copied multicast packet to the unicast packet buffer queue according to the priority, and then sends the packet to the downlink module through the local forwarding channel according to the QoS setting of the multicast member port. And the downlink module sends the duplicated multicast packet to the multicast member port included in the physical port of the egress traffic manager.

According to another aspect of the present invention, an embodiment of the present invention further provides an egress traffic manager, including an uplink module, a downlink module, and a physical port for communicating with the outside, where the downlink module is configured with a loop port, where: The downlink module is configured to receive the multicast packet sent by the switching network by using the cyclic port, and send the multicast packet to the uplink module.

The uplink module is configured to search for a multicast member port corresponding to the multicast packet in the first uplink multicast replication table according to the group identifier carried in the multicast packet, and calculate the multicast packet corresponding to the multicast packet The number of multicast member ports belonging to the physical port of the egress traffic manager in the multicast member port, and the multicast report according to the number of multicast member ports included in the physical port of the egress traffic manager The file is copied, so that the number of the multicast packets after the replication is the same as the number of the multicast member ports included in the physical port of the egress traffic manager; The unicast packet cache queue is added to the unicast packet buffer queue and sent to the downlink module through the local forwarding channel after being scheduled according to the multicast member port QoS.

The downlink module is further configured to send the copied multicast "3" message to the multicast member port included in the physical port of the egress traffic manager.

According to still another aspect of the present invention, an embodiment of the present invention further provides a multicast replication system, including an egress traffic manager, an ingress traffic manager, and an egress traffic manager connected to the The switching network between the ingress traffic managers.

According to the multicast replication method, device, and system provided by the embodiment of the present invention, after the downlink module of the egress traffic manager accesses the multicast packet, the multicast packet is looped back to itself through the loop port set by itself. The uplink module, in the uplink module of the egress traffic manager, implements multicast packet replication for the multicast member port, and the copied multicast packet enters the unicast packet buffer queue according to the priority, according to the multicast member. After the port QoS is scheduled, it is sent to the downlink module through the local forwarding channel, and is sent out by the downstream module through the multicast member port included in the egress traffic manager. Therefore, the multicast member port QoS can be scheduled, which can be effectively implemented. Provide QoS guarantees under the condition of saving the bandwidth of the switching network. DRAWINGS

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will be true. The drawings used in the examples or the description of the prior art are briefly introduced. It is obvious that the drawings in the following description are only some embodiments of the present invention, and are not creative to those skilled in the art. Other drawings can also be obtained from these drawings on the premise of labor.

1 is a schematic flowchart of a first multicast replication method in the prior art;

2 is a schematic flowchart of a second multicast replication method in the prior art;

3 is a schematic flowchart of a third multicast replication method in the prior art;

4 is a flowchart of a multicast replication method according to a first embodiment of the present invention;

5 is a system architecture diagram of a multicast replication method to which a first embodiment of the present invention is applied;

Figure 6 is a schematic view showing the structure of an outlet TM according to a second embodiment of the present invention. detailed description

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

Embodiment 1

FIG. 4 is a flowchart of a multicast copying method according to a first embodiment of the present invention. As shown in Figure 4, the multicast replication method includes the following steps:

Step S101: The downlink module of the egress TM receives the multicast packet sent by the switching network through the cyclic port, and sends the multicast packet to the uplink module.

The egress TM includes an uplink module, a downlink module, and a physical port for communicating with the external, the cyclic port is set on the downlink module, and the cyclic port passes the internal loopback channel and the uplink module disposed between the cyclic port and the uplink module. When the loop port is used as the communication interface between the egress TM and the switching network, after receiving the multicast message from the switching network, the loop port directly sends the received multicast packet from the downlink module to the upper loop through the internal loopback channel. Line module. The uplink module stores the received packets in different address spaces in order according to the difference between the destination port and the priority. Step S102: After receiving the multicast packet, the uplink module of the egress TM searches for the multicast packet in the first uplink multicast replication table according to the group identifier carried in the multicast packet. a multicast member port, where the first uplink multicast replication table stores a group identifier of the multicast packet and a multicast member port corresponding to the group identifier, and the format of the uplink multicast replication table is, for example, Table 1;

Table 1

Step S103, the uplink module of the egress TM calculates the number of multicast member ports of the multicast member port corresponding to the egress traffic manager in the multicast member port corresponding to the multicast packet, and according to the egress traffic manager The number of multicast member ports included in the physical port is copied to the multicast packet, so that the number of the copied multicast packets and the multicast member port included in the physical port of the egress TM are The same number;

Step S104, the egress ME adds the copied multicast packet to the unicast packet buffer queue according to the priority, and then sends the packet to the downlink module through the local forwarding channel according to the QoS setting of the multicast member port. The downlink module sends the copied multicast packet to the multicast member port included in the physical port of the egress traffic manager. The local forwarding channel is set between the upstream module and the downlink module of the egress TM, and the to-be-sent packet buffered by the egress module is forwarded to the downlink module through the local forwarding channel.

FIG. 5 is a system architecture diagram of a multicast replication method to which the first embodiment of the present invention is applied. As shown in FIG. 5, the system includes a plurality of line cards and a switching network, wherein each line card can include one or more TMs. Each The line card also includes a network processor (not shown in Figure 5), and the line card is connected to the switching network. Where TM is a data frame that restores a physical signal on a transmission line (eg, fiber, cable, or wireless channel, etc.) to an agreed format for further processing by the network processor, or completes a reverse process, ie, the data is processed by the network processor The frame is processed, and the processed data frame is converted by the TM into a physical signal on the transmission line and transmitted to the transmission line. For the multicast packet, the network processor identifies the source information, the destination information, and the preference information of the multicast packet according to the agreed format, and then modifies the packet format according to the current forwarding rule, sends the packet format to the TM, and obtains the multicast from the network processor. The packet sent by the multicast packet to the multicast member port is the entry TM of the multicast packet, and the TM of the multicast member port of the multicast packet is the multicast packet. Text ExportTM. The multicast copying method of the first embodiment of the present invention described above is described from the perspective of the ExitTM.

Specifically, TM-4 is an exit TM of the multicast message shown in FIG. 5, and the multicast copy method of the first embodiment of the present invention will be described below by taking the TM-4 shown in FIG. 5 as an example. The TM-4 receives the multicast packet from the switching network through its own cyclic port C, and the cyclic port C directly loops the received multicast packet back to the uplink module of the TM-4 via the internal loopback channel. Afterwards, the uplink module of the TM-4 queries the entry corresponding to the group ID in the uplink multicast replication table according to the group identifier (group ID) carried in the header of the multicast packet, specifically, corresponding to the group ID. The entries include all multicast member ports corresponding to the multicast message.

Next, the uplink module of the TM-4 filters out the multicast member ports included in the TM-4 from all the multicast member ports corresponding to the multicast packet, and according to the number of multicast member ports included in the TM-4. To copy the multicast packets, the number of copies to be added is "the number of multicast member ports included in the TM-4 minus 1", that is, the number of duplicated multicast packets and the TM-4 are ensured. The number of multicast member ports is the same. For example, TM-4 can query the uplink multicast replication table of the device to learn that all multicast member ports of the multicast packets are four, and two of them are the ports of the TM-4. At this time, the TM-4 only needs to copy one multicast packet to obtain two multicast packets corresponding to two ports.

The packet sent by the uplink module of the TM is the switching network. Therefore, after the uplink module of the TM-4 completes the replication of the local multicast member port, it needs to be copied through the local forwarding channel. The multicast packet is forwarded to the local downlink module, and the multicast packet is sent to the network processor through the multicast member port included in the TM-4 in the manner that one multicast packet corresponds to one port. That is, the downlink module stores the multicast packet in the cache of the local multicast member port and sends it to the network processor for further processing to the network processor.

According to the multicast replication method of the foregoing embodiment, after the egress packet receives the multicast packet, the egress TM directly loops back to the local uplink module through the internal loop port, and implements the replication of the multicast message in the TM. The local forwarding channel forwards the copied multicast "3" to the downstream module of the egress TM. On the one hand, the copied multicast packet is queued to the normal unicast queue according to the priority, and the unicast packet is shared. Scheduling, accepting the unified scheduling of the downlink port, its QoS performance can be fully guaranteed; on the other hand, because the loopback and replication of the packet are implemented in the same TM, the multicast packet is only used by the downstream port. The bandwidth of the internal loopback channel is occupied in the process of returning to the uplink module, and the bandwidth of the local forwarding channel is occupied by the uplink module in the process of forwarding the copied multicast packet from the uplink module to the downlink module, without consuming the bandwidth of the switching network. The switching network forwards the bandwidth, which effectively saves the bandwidth of the switching network, thereby improving the replication efficiency.

Further, in the multicast replication method of the foregoing embodiment, before the egress TM receives the multicast packet sent by the switching network through the re-circulation port, the method further includes:

The ingress TM receives the multicast packet sent by the network processor, and searches for the egress TM corresponding to the multicast packet in the second uplink multicast replication table according to the group identifier carried in the multicast packet, where the second uplink group The group identifier of the multicast packet stored in the broadcast replication table and the identifier of the exit TM corresponding to the group identifier are only used;

The ingress TM copies the multicast packet, so that the number of the copied multicast packets is the same as the number of the egress TMs;

The ingress TM sends the multicast message to the egress TM through the switching network.

Specifically, in the multicast replication method of the foregoing embodiment, the egress TM completes the multicast replication for the local multicast member port in the chip, where the ingress TM completes the different TM for the multicast member port. Multicast packet replication. TM-0 is the entry of the multicast message shown in Figure 5. The TM, TM-0, after receiving the multicast packet from the network processor, searches the local uplink multicast replication table according to the group ID carried in the multicast packet to learn the egress TM corresponding to the multicast packet. Taking the situation shown in Figure 5 as an example, the egress TM corresponding to the multicast packet includes TM-4 and TM-3. At this time, TM-0 needs to copy the multicast and subtext once (not shown in Figure 5). Shown) to obtain two multicast messages that can be sent to ΤΜ-4 and ΤΜ-3 via the switching network, respectively.

Further, in the multicast replication method of the foregoing embodiment, before the egress port receives the multicast packet sent by the switching network through the re-circulation port, the method further includes:

The switching network receives the multicast packet sent by the ingress port, and searches for the egress packet corresponding to the multicast packet in the multicast replication table according to the group ID carried in the multicast packet, where the multicast replication table stores the a group identifier of the multicast packet and an identifier of the exit TM corresponding to the group identifier;

The switching network replicates the multicast packets so that the number of replicated multicast packets is the same as the number of egress TMs.

The switching network sends a multicast message to the egress TM, that is, sends a multicast message to each egress TM.

Specifically, in the multicast replication method of the foregoing embodiment, the egress TM completes the multicast packet replication for the local multicast member port in the chip, where the switching network completes different TMs for the multicast member port. Multicast packet replication. TM-0 is the ingress TM of the multicast packet shown in FIG. 5, that is, after the switching network receives the multicast packet from the TM-0, the switching network searches for the locally stored multicast replication table according to the group ID carried in the multicast packet. To know the egress TM corresponding to the multicast packet. As shown in Figure 5, the egress TM corresponding to the multicast packet includes TM-4 and TM-3. In this case, the switching network needs to copy the multicast packet (as shown in Figure 5). Show), to obtain two multicast messages that can be sent to ΤΜ-4 and ΤΜ-3 respectively.

When the above-mentioned scheme for performing multicast packet replication for different egresses by the switching network is used, since the upstream traffic of the ingress to the switching network has only one multicast packet traffic, it is different for the completion of the portal port. The scheme of multicast 艮复制复制复制复制复制复制 , 能够能够能够复制复制复制复制复制复制复制复制复制复制复制复制复制 Further, in the multicast replication method of the foregoing embodiment, the multicast packet sent to the uplink module carries an identifier for identifying a cyclic port of the multicast packet from the egress TM.

Specifically, after receiving the multicast packet sent by the switching network through the cyclic port, the egress TM configures an identifier for the multicast packet before the multicast packet is sent to the uplink module, so that the multicast packet can be It is different from the multicast packet received from the external network processor when the egress TM is used as the ingress traffic manager of other multicast packets. That is, after the uplink module of the egress TM detects that the received multicast packet carries the identifier of the re-circulating port for identifying the multicast packet from the egress TM, step S102 to step S104 in the above embodiment are executed.

Embodiment 2

Figure 6 is a schematic view showing the structure of an outlet TM according to a second embodiment of the present invention. As shown in FIG. 6, the egress TM includes an uplink module 202 and a downlink module 201. The egress TM further includes a physical port (not shown) for communicating with the outside, and the downlink module is provided with a cyclic port, wherein The downlink module 201 is configured to receive the multicast packet sent by the switching network through the cyclic port, and send the multicast packet to the uplink module 202;

The uplink module 202 is configured to search for a multicast member port corresponding to the multicast packet in the first uplink multicast replication table according to the group identifier carried in the multicast packet, and calculate a corresponding corresponding to the multicast packet. The number of multicast member ports of the multicast member port that belong to the physical port of the egress traffic manager, and the multicast packet according to the number of multicast member ports included in the physical port of the egress traffic manager Performing the replication, so that the number of the multicast packets after the replication is the same as the number of the multicast member ports included in the physical port of the egress traffic manager; After being added to the unicast packet cache queue, it is sent to the downlink module 201 through the local forwarding channel according to the multicast member port QoS setting.

The downlink module 201 is further configured to send the copied multicast text to the multicast member port included in the physical port of the egress traffic manager. The process of performing the multicast replication in the egress TM of the foregoing embodiment is the same as the multicast replication method in the first embodiment of the present invention, and therefore is not described herein again. According to the egress TM of the foregoing embodiment, after receiving the multicast packet, the downlink module directly loops back to the uplink module through the internal loop port, and the uplink module searches for the uplink multicast replication table to implement in the uplink module of the TM. The multicast packet is copied, and the copied multicast packet is forwarded to the downlink module of the local TM through the local forwarding channel. On the one hand, the copied multicast packet enters the unicast packet buffer queue according to the priority. The hop scheduling is shared with the unicast packet, and the unicast scheduling of the downlink port is accepted. The QoS performance of the downlink port can be fully guaranteed. On the other hand, the loopback and replication of the packet are implemented in the same TM. The process of occupying the bandwidth of the internal loopback channel and occupying the bandwidth of the local forwarding channel during the process of forwarding the copied multicast packet from the upstream to the downlink does not consume the bandwidth of the switching network and the forwarding bandwidth of the switching network, thereby effectively saving the bandwidth. Switch network bandwidth, which improves replication efficiency.

Further, in the egress TM of the foregoing embodiment, the downlink module is further configured to add, in the multicast packet sent to the uplink module, an identifier for identifying the cyclic port of the multicast packet from the egress traffic manager.

Specifically, after receiving the multicast packet sent by the switching network through the cyclic port, the downlink module configures an identifier for the multicast packet before the multicast packet is sent to the uplink module, so that the multicast packet can be It is different from the multicast message received by the external network processor when the egress TM is used as the ingress traffic manager of other multicast messages. That is, the uplink module performs the corresponding operation if it detects that the received multicast packet carries an identifier for identifying the multicast packet from the egress port of the egress port. With this arrangement, the egress TM can be used as an entry point for other multicast messages.

Embodiment 3

A third embodiment of the present invention provides a multicast replication system including an egress TM, an entry TM, and a switching network connected between the egress TM and the portal TM provided by the second embodiment of the present invention.

In the multicast replication system of the foregoing embodiment, the chip-level multicast packet replication for different egress packets corresponding to the multicast packet is completed by the ingress TM, and the copied packet is sent to each egress TM through the switching network. And then the TM completes the port for the multicast member port corresponding to the multicast message. The multicast packets are transmitted and sent out through the local multicast member port. The multicast packets are sent from the external network processor to the switching network. The multicast packets are sent to the switching network. The chip-level multicast packets of the different egress TMs are copied, and the copied packets are respectively sent to the egress TMs, and the TM completes the port-level group for the multicast member ports corresponding to the multicast packets. The broadcast message is copied and sent out via the local multicast member port. Preferably, the chip-level multicast copy of the different egress TM corresponding to the multicast message is completed by the switching network.

According to the multicast replication system of the foregoing embodiment, on the one hand, the multicast packet copied by the egress TM can be queued to the common unicast queue, and the unicast packet is shared with the queue scheduling, and the unified scheduling of the downlink port is accepted, and the Qos performance is obtained. On the other hand, since the loopback and replication of the packets completed in the egress TM are implemented in the same TM, the internal loopback channel bandwidth is occupied only during the downlink loopback to the uplink queue, and During the process of forwarding the copied multicast packet from the upstream to the downlink, the bandwidth of the local forwarding channel is occupied without consuming the bandwidth of the switching network and the forwarding bandwidth of the switching network. Moreover, when the chip-level multicast of the different egress TMs corresponding to the multicast message is copied by the switching network, only one multicast packet traffic is generated between the ingress TM and the switching network, so further The bandwidth of the switching network is saved, and the replication efficiency is greatly improved.

It will be understood by those skilled in the art that all or part of the steps of implementing the foregoing embodiments may be performed by hardware related to program instructions. The foregoing program may be stored in a computer readable storage medium, and when executed, the program includes The foregoing steps of the method embodiment; and the foregoing storage medium includes: a medium that can store program codes, such as a ROM, a RAM, a magnetic disk, or an optical disk.

It should be noted that the above embodiments are only for explaining the technical solutions of the present invention, and are not intended to be limiting; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those skilled in the art that: The technical solutions described in the foregoing embodiments are modified, or some of the technical features are equivalently replaced. The modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

Rights request

A multicast replication method, comprising:

The uplink module of the egress traffic manager adds the copied multicast packet to the unicast packet buffer queue according to the priority, and then sends the packet to the downlink through the local forwarding channel according to the multicast member port quality of service QoS setting. And sending, by the downlink module, the copied multicast packet to the multicast member port included in the physical port of the egress traffic manager.

The multicast replication method according to claim 1, wherein before the egress traffic manager receives the multicast packet sent by the switching network, the egress traffic manager further includes:

The ingress traffic manager receives the multicast packet sent by the network processor, and searches for the egress corresponding to the multicast packet in the second uplink multicast replication table according to the group identifier carried in the multicast packet. a traffic manager, where the second uplink multicast replication table stores a group identifier of the multicast packet and an identifier of an egress traffic manager corresponding to the group identifier;

The ingress traffic manager replicates the multicast packet, so that the number of the copied multicast packets is the same as the number of the egress traffic manager;

The ingress traffic manager sends the group to the egress traffic manager through the switching network Broadcast message.

The multicast replication method according to claim 1, wherein before the egress traffic manager receives the multicast packet sent by the switching network through the cyclic port, the method further includes:

The switching network receives the multicast packet sent by the ingress traffic manager, and searches for the egress traffic manager corresponding to the multicast packet in the multicast replication table according to the group identifier carried in the multicast packet. And storing, in the multicast replication table, a group identifier of the multicast packet and an identifier of an egress traffic manager corresponding to the group identifier;

The switching network replicates the multicast packet, so that the number of the copied multicast packets is the same as the number of the egress traffic manager;

The switching network sends the multicast packet to the egress traffic manager.

The multicast replication method according to claim 1 or 2 or 3, wherein the multicast packet sent to the uplink module carries a identifier for identifying that the multicast packet is from the egress. The identity of the traffic manager's circular port.

An export traffic manager, comprising: an uplink module, a downlink module, and a physical port for communicating with the outside, wherein the downlink module is provided with a cyclic port, where:

The downlink module is configured to receive the multicast packet sent by the switching network by using the cyclic port, and send the multicast packet to the uplink module.

The downlink module is further configured to pass the copied multicast message to the egress traffic manager. The multicast member port included in the physical port is sent out.

The egress traffic manager according to claim 5, wherein the downlink module is further configured to add, in the multicast packet sent to the uplink module, the identifier to identify the multicast packet from the egress traffic manager. The identity of the loop port.

A multicast replication system, comprising: an egress traffic manager according to claim 5 or 6, an ingress traffic manager, and a connection between the egress traffic manager and the ingress traffic manager Switching network.