CN117880223A - Two-layer annular reliable switch clustering method for large-scale IP multicast service - Google Patents

Abstract

The invention discloses a two-layer annular reliable exchanger clustering method of large-scale IP multicast service, comprising the following steps: interconnecting switch 1, switch 2, switch 3 and switch 4 into 1 ring network, thereby forming a ring cluster; configuring adjacent switches 1 and 2 as an aggregation switch; the switch 3 and the switch 4 are set as a normal switch; dividing ports of an aggregation switch, a switch 3 and a switch 4 into an inter-machine interconnection port and a common end equipment access port; adopting modified IGMP snoring configuration to the ring cluster to obtain a mapping relation table of the IP multicast service frame and the destination port; when receiving the IP multicast service frame, the exchanger finds a destination port corresponding to the IP multicast service frame received by the exchanger according to the mapping relation table, and sends the IP multicast service frame to the corresponding destination port, and meanwhile forwards the IP multicast service frame to the adjacent exchanger, thereby realizing the two-layer forwarding of the IP multicast service.

Description

Two-layer annular reliable switch clustering method for large-scale IP multicast service

Technical Field

The invention belongs to the field of switch networks, and particularly relates to a two-layer annular reliable switch clustering method for large-scale IP multicast service.

Background

The processing of large capacity and high speed data streams has become increasingly important and urgent in the internet and communication arts today. From network communication to cloud computing, from big data analysis to real-time video transmission, the ability to efficiently handle large data streams is needed. The switch is used as a key device in the network, is a key node for forwarding network data for all terminal devices, and has the performance and reliability which are important for the terminal devices which need to forward the service for the whole network. Meanwhile, as the complexity of the service environment increases, the number of the end devices which are required to be accessed by a single switch, and the number of containers which are downloaded by the single end device are increased rapidly. Large-scale IP multicast traffic is becoming more and more common. The IP multicast technology provides an efficient data transmission mode, and can simultaneously transmit a large amount of audio and video streams, real-time data and other multimedia contents in a wide area network and a local area network. However, due to the increasing amount of multicast data and the demand for network capacity, conventional network architecture and switch designs have become no longer suitable for handling large-scale IP multicast traffic. Therefore, a new switch clustering method is needed, which can meet the requirements of modern large-scale IP multicast service, and also provide reliability requirements for forwarding data for end devices.

Currently, many large organizations and enterprises use a two-tier ring topology in their networks to provide high availability and redundant connections. However, in the conventional two-layer ring topology, such a topology cannot effectively handle large-scale IP multicast traffic due to backpressure problems between switches and cyclic forwarding of data packets. In multicast transmission, duplication and delivery of data packets often results in increased network congestion and delay, thereby degrading system performance and user experience. To support the large-scale number of end devices and reliability of end device data in the current switch network, the accessed switch devices must have sufficient resources. Because the current stage of using the FPGA to realize the switch design is more convenient than a chip, a plurality of manufacturers use the FPGA to realize the switch design, but the problem of the FPGA is that logic resources and storage resources are limited, and a single FPGA board card cannot support the switching requirement of large-scale equipment access.

Through the above analysis, the problems and defects existing in the prior art are as follows:

(1) The number of ports of the existing single switch is too small, and large-scale access of terminal equipment is difficult to carry out;

(2) The single switch board card can not store enough multicast list items under the limitation of FPGA internal resources, especially when IP multicast service is carried out;

(3) The existing switch network cannot improve the data forwarding reliability of the access device, and even cannot backup a single data stream. Once the network link is disconnected, the end device data may not be transmitted;

(4) The conventional switch can form cascade connection, but the cascade topology is not flexible enough and can not improve the reliability of network links, and the end devices with different service requirements can not be treated differently.

Disclosure of Invention

In order to solve the problems in the prior art, the invention provides a two-layer annular reliable switch clustering method for large-scale IP multicast service. The technical problems to be solved by the invention are realized by the following technical scheme:

a two-layer annular reliable switch clustering method for large-scale IP multicast service comprises the following steps:

interconnecting switch 1, switch 2, switch 3 and switch 4 into 1 ring network, thereby forming a ring cluster;

configuring the adjacent switch 1 and the adjacent switch 2 as an aggregation switch; the aggregation switch is used for accessing user equipment and special requirement equipment;

setting the switch 3 and the switch 4 as a normal switch; the switch 3 and the switch 4 are both used for accessing network end equipment and user end equipment;

dividing the ports of the aggregation switch, the switch 3 and the switch 4 into inter-machine interconnection ports and common end equipment access ports;

adopting modified IGMP snoring configuration to the annular cluster to obtain a mapping relation table of an IP multicast service frame and a destination port;

when receiving the IP multicast service frame, the switch in the ring cluster finds a destination port corresponding to the IP multicast service frame received by the switch according to the mapping relation table, and sends the IP multicast service frame to the corresponding destination port, and meanwhile forwards the IP multicast service frame to an adjacent switch, so that the two-layer forwarding of the IP multicast service is realized.

In one embodiment of the present invention, the CPUs of the switch 1 and the switch 2 are configured with aggregation group information; the aggregation group information includes: group number and priority information; the group number is an identity in an aggregation group formed by the switch 1 and the switch 2; the priority information indicates a priority level of the switch operation.

In one embodiment of the present invention, the adjacent switch 1 and the adjacent switch 2 are configured as an aggregation switch, including:

configuring the aggregation group information by the CPU of the switch 1 and the CPU of the switch 2, and configuring the aggregation group information to respective hardware FPGA boards through PIO;

the hardware FPGA board cards of the switch 1 and the switch 2 carry out framing encapsulation on the received aggregation group information after the configuration respectively, and broadcast a local port as an aggregation group hello message so as to send the aggregation group hello message to an adjacent switch;

after the adjacent exchanger receives the corresponding aggregation group hello message, inquiring whether the adjacent exchanger meets the preset condition is as follows: the group number of the aggregation group is the same as the group number of the aggregation group in the received aggregation group hello message; if yes, determining a group of switches for sending and receiving the aggregation group hello message as the switches to be paired; the ports of the switches to be paired, which are respectively responsible for sending and receiving the aggregation group hello messages, are connected with each other and are determined to be the interconnection ports in the groups, so that an aggregation switch is formed;

and when the aggregation is completed, the switch receiving the corresponding aggregation group hello message compares the priority information in the received aggregation group hello message with the own priority information, and confirms the master device and the slave device.

In one embodiment of the present invention, the switch receiving the corresponding aggregation group hello packet compares priority information in the corresponding aggregation group hello packet with own priority information, and confirms the master device and the slave device, including:

the exchanger receiving the aggregation group hello message judges whether the priority information in the received aggregation group hello message is greater than the priority information of the exchanger per se;

if yes, determining a switch for sending the aggregation group hello message as a master device, and determining a switch for receiving the aggregation group hello message as a slave device;

if not, determining the exchanger for receiving the aggregation group hello message as a master device, and determining the exchanger for sending the aggregation group hello message as a slave device.

In one embodiment of the invention, the link of the intra-group interconnect port is a peer link.

In one embodiment of the present invention, the dividing the ports of the aggregation switch, the switch 3 and the switch 4 into the division modes of the inter-machine interconnection port and the normal end device access port includes:

manual division mode and automatic recognition division mode.

In one embodiment of the present invention, the manual partition mode operation includes:

after receiving a setting instruction of a user, a CPU of the current switch transmits an inter-machine interconnection port number of the current switch to a hardware FPGA board card of the current switch, and the hardware FPGA board card of the current switch is maintained.

In one embodiment of the present invention, the operation of automatically identifying the division pattern includes:

the CPU of the current switch issues a port discovery instruction to a hardware FPGA board card of the current switch, the hardware FPGA board card of the current switch carries out framing after perceiving the port discovery instruction, the information after framing comprises a switch number of the current switch, and the current switch forwards the information after framing in a broadcasting mode;

when the port of the exchanger receives the information carrying the framing of the non-self exchanger number, the port receiving the information after framing is set as an inter-machine interconnection port, and the other ports are set as common terminal equipment access ports;

when the network access environment changes, the port division is carried out again.

In one embodiment of the present invention, a working process of obtaining a mapping relation table of an IP multicast service frame and a destination port by adopting a modified IGMP snooping configuration for the ring cluster includes:

1 exchanger is selected in the ring cluster, and a common group query request is issued through the CPU of the selected exchanger;

after receiving the normal group query request, the hardware FPGA board card of the selected switch carries out framing of a normal group query message, and sends the framed normal group query message to all ports;

after receiving the ordinary group query message after framing, the user side equipment supporting the IGMP protocol responds to generate a group adding message and sends the group adding message to a switch connected with the user side equipment, wherein the group adding message comprises the IP multicast address of the user side equipment supporting the IGMP protocol; the exchanger connected with the user side equipment receives the grouping message and sends the grouping message to the CPU of the exchanger; and the CPU of the CPU analyzes the grouping message, extracts the IP multicast address in the grouping message, and forms a mapping relation table of the IP multicast address in the grouping message and the port where the IP multicast address is located.

In one embodiment of the present invention, after receiving an IP multicast service frame, a switch in the ring cluster searches the mapping relation table to find a destination port, and forwards the IP multicast service frame to the destination port, so as to implement a working process of forwarding the IP multicast service by the ring cluster, where the working process includes:

after receiving the IP multicast service frame, the exchanger in the ring cluster analyzes the IP multicast address in the IP multicast service frame, queries the mapping relation table according to the analyzed IP multicast address to obtain a corresponding destination port, and sends the received IP multicast service frame to the corresponding destination port; and forwarding the received IP multicast service frame to two adjacent switches through the inter-machine interconnection port.

The invention has the beneficial effects that:

compared with the prior art, the embodiment of the invention has the following advantages:

(1) The cluster structure formed by 4 switches is adopted, so that the access scale is larger, more end devices are supported compared with the common switches, and the switches can still be added into the cluster model for ring connection so as to enhance expansibility;

(2) Under the condition of limited FPGA resources, a larger forwarding table item can be maintained in the cluster, so that the forwarding capacity and the processing speed of the IP multicast service are improved;

(3) When one link of the user side equipment fails, the aggregation switch selects other links to forward, and two switches in the aggregation switch are redundant, so that the reliability and redundancy of the equipment are improved;

(4) By means of annular cluster design, the switch 3 and the switch 4 are respectively connected with the network side, and when one inter-machine link of the switch fails or the network side link fails, information forwarding is guaranteed, so that higher reliability and load balancing are achieved.

Drawings

Fig. 1 is a schematic flow chart of a method for clustering two-layer ring-shaped reliable switches of a large-scale IP multicast service according to an embodiment of the present invention;

fig. 2 is a schematic diagram of a two-layer ring-shaped reliable switch clustering method for large-scale IP multicast service according to an embodiment of the present invention;

fig. 3 is a schematic structural diagram of an aggregation switch of a two-layer ring-shaped reliable switch clustering method for large-scale IP multicast service according to an embodiment of the present invention;

fig. 4 is a flowchart of configuring an aggregation switch in a two-layer ring-shaped reliable switch clustering method for a large-scale IP multicast service according to an embodiment of the present invention;

fig. 5 is a flow chart of port division in a method for clustering two-layer ring-shaped reliable switches of a large-scale IP multicast service according to an embodiment of the present invention;

fig. 6 is a flowchart of ring-shaped cluster IGMP snooping configuration in a two-layer ring-shaped reliable switch clustering method for large-scale IP multicast service according to an embodiment of the present invention;

fig. 7 is a schematic diagram of forwarding a data stream of a user side link failure in a method for clustering two-layer ring-shaped reliable switches of a large-scale IP multicast service according to an embodiment of the present invention;

fig. 8 is a schematic diagram of forwarding a data flow of an inter-machine link failure in a method for clustering two-layer ring-shaped reliable switches of a large-scale IP multicast service according to an embodiment of the present invention;

fig. 9 is a schematic diagram of forwarding a data flow of a network side link failure in a method for clustering a two-layer ring-shaped reliable switch for a large-scale IP multicast service according to an embodiment of the present invention.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

The embodiment of the invention provides a two-layer annular reliable switch clustering method for large-scale IP multicast service, which can comprise the following steps as shown in figure 1:

s1, interconnecting a switch 1, a switch 2, a switch 3 and a switch 4 into 1 annular network, thereby forming an annular cluster;

s2, configuring adjacent switches 1 and 2 as an aggregation switch; the aggregation switch is used for accessing the user terminal equipment and the special requirement equipment;

s3, setting the switch 3 and the switch 4 as a common switch; the switch 3 and the switch 4 are used for accessing network terminal equipment and user terminal equipment;

s4, dividing ports of the aggregation switch, the switch 3 and the switch 4 into inter-machine interconnection ports and common terminal equipment access ports;

s5, adopting modified IGMP snoring configuration to the annular cluster to obtain a mapping relation table of the IP multicast service frame and the destination port;

and S6, when the exchanger in the annular cluster receives the IP multicast service frame, finding a destination port corresponding to the IP multicast service frame received by the exchanger according to the mapping relation table, sending the IP multicast service frame to the corresponding destination port, and forwarding the IP multicast service frame to an adjacent exchanger to realize the two-layer forwarding of the IP multicast service.

Referring to fig. 2, referring to a schematic diagram of a method for clustering two-layer ring-shaped reliable switches of a large-scale IP multicast service provided by the embodiment of the present invention, a ring-shaped cluster model provided by the embodiment of the present invention is composed of 4 switches, each switch is made of an independent FPGA board, and the 4 switches form a ring-shaped network through interconnection, so as to form a ring-shaped cluster. The 4 exchanges are distinguished according to different configurations and functions, and the exchange 1 and the exchange 2 are externally presented as 1 exchange through aggregation configuration, which is called an aggregation exchange. The aggregation switch is used for accessing the user terminal equipment and the special requirement equipment. Special requirements devices may include devices that are highly reliable and load demanding. The other switches 3 and 4 are not configured in an aggregation manner, and are called normal switches, and are used for accessing network side equipment and user side equipment.

For S2, referring to fig. 3, in the embodiment of the present invention, an aggregation group formed by two switches exists, and configuration of the aggregation group needs to be completed in the early stage of networking, typically, two adjacent switches in the ring cluster form an aggregation group, and for an accessed device, two switches in the aggregation group are virtualized as one switch, so that reliability is improved from a link level to a device level.

In the initial stage of forming the ring cluster through the link, there is no aggregation group, and the configuration of the aggregation group needs to be performed at the beginning, and a flow chart of configuring the aggregation switch is shown in fig. 4, where the configuration flow of the aggregation switch is as follows:

specifically, configuring the adjacent switch 1 and switch 2 as an aggregation switch may include:

s21, configuring aggregation group information by the CPU of the exchanger 1 and the exchanger 2, and configuring the aggregation group information to respective hardware FPGA board cards through PIO;

the CPUs of the switch 1 and the switch 2 are configured with aggregation group information; aggregation group information, may include: group number and priority information; the group number is the identity in the aggregation group formed by the exchanger 1 and the exchanger 2; the priority information indicates the priority of the switch operation.

S22, the hardware FPGA boards of the exchanger 1 and the exchanger 2 carry out framing encapsulation on the received aggregation group information after the configuration respectively, and broadcast a local port as an aggregation group hello message so as to send the aggregation group hello message to an adjacent exchanger;

s23, after the adjacent exchanger receives the corresponding aggregation group hello message, inquiring whether the adjacent exchanger meets the preset condition or not is as follows: the group number of the aggregation group is the same as the group number of the aggregation group in the received aggregation group hello message; if yes, determining a group of switches for sending and receiving the aggregation group hello message as the switches to be paired; the method comprises the steps that ports of switches to be paired, which are respectively responsible for sending and receiving an aggregation group hello message, are connected with each other, and are determined to be interconnection ports in the group, so that an aggregation switch is formed;

and S24, when the aggregation is completed, the switch receiving the corresponding aggregation group hello message compares the priority information in the received aggregation group hello message with the own priority information, and confirms the master device and the slave device.

For S24, it may include:

the exchanger receiving the aggregation group hello message judges whether the priority information in the received aggregation group hello message is greater than the priority information of the exchanger per se;

if yes, determining the exchanger for transmitting the aggregation group hello message as a master device and the exchanger for receiving the aggregation group hello message as a slave device;

if not, determining the exchanger for receiving the aggregation group hello message as a master device and the exchanger for transmitting the aggregation group hello message as a slave device.

It should be noted that the same ue may access different switches in the aggregation group at the same time, and the two access links are equivalent, because the switch in the aggregation group appears to the ue as one switch.

The links of the inter-connected ports in the group are peer link links. After the pairing of the aggregation group is completed, the peer link will not forward data in general, and is only used for transmitting synchronization information and aggregation group configuration information, for example: multicast forwarding table entry information, MAC table entry information, etc., switches within an aggregation group share such information for forwarding.

After the configuration of the aggregation group is completed, the ports with the same port numbers of the switches in the aggregation group are set as a group of aggregation member ports, and the user side equipment can realize the function of link aggregation by simultaneously accessing the group of aggregation member ports. Devices need to support the LACP protocol to access the aggregation group and use its aggregation functionality. After the master device completes pairing, LACP negotiation is performed, the slave device can modify the MAC addresses of all aggregation member ports on the master device to be the same as the system MAC of the master device, so that when LACP negotiation is performed, the same system ID is contained in LACP PDUs sent by the master device and the slave device, and user side devices supporting the LACP protocol feel that negotiation is performed with the same device, thereby achieving the purpose of link aggregation across devices.

For S3, switch 3 and switch 4 are both used to access network side devices and customer side devices.

After the switch ring cluster is configured, the inter-extension interconnection ports and the common equipment access ports are also required to be divided. The embodiment of the invention provides a flow chart of port division of a method for clustering two-layer annular reliable switches of large-scale IP multicast service, and please refer to fig. 5.

For S4, dividing the ports of the aggregation switch, switch 3 and switch 4 into a division mode of an inter-machine interconnection port and a normal end device access port includes:

manual division mode and automatic recognition division mode.

S41, a working process of a manual dividing mode comprises the following steps:

after receiving a setting instruction of a user, a CPU of the current switch transmits an inter-machine interconnection port number of the current switch to a hardware FPGA board card of the current switch, and the hardware FPGA board card of the current switch is maintained.

S42, automatically identifying the working process of the division mode, comprising:

the CPU of the current switch issues port discovery instructions to the hardware FPGA board card of the current switch, the hardware FPGA board card of the current switch frames after sensing the port discovery instructions, the information after framing comprises the switch number of the current switch, and the current switch forwards the information after framing in a broadcasting mode;

when the port of the exchanger receives the information carrying the framing of the non-self exchanger number, the port receiving the information after framing is set as an inter-machine interconnection port, and the other ports are set as common terminal equipment access ports;

when the network access environment changes, the port division is carried out again.

It should be noted that taking the overall structure shown in fig. 2 as an example, both the switch 3 and the switch 4 as the normal switches will generate the port discovery frame and broadcast it; switch 1 and switch 2 as members of the aggregate group would likewise generate respective port discovery frames, but they would mask out intra-group interconnect ports (port 2 of switch 1 and port2 of switch 2) for broadcast. Port discovery frames are forwarded in the network device, for example, switch 3, and port1 and port2 of switch 3 receive port discovery frames from switch 1 and switch 4, respectively, so that their own port1 and port2 are considered as inter-machine interconnect ports.

For S5, the switch cluster supports two-layer forwarding of the IP multicast service, but needs to adopt a modified IGMP snooping configuration, and maintains an IP multicast forwarding table in the switch.

Specifically, the working process of obtaining the mapping relation table of the IP multicast service frame and the destination port by adopting the modified IGMP snooping configuration for the ring cluster may include:

s51, 1 exchanger is selected in the ring cluster, and a common group query request is issued through a CPU of the selected exchanger;

s52, after receiving the normal group query request, the hardware FPGA board card of the selected switch carries out framing of the normal group query message, and sends the framed normal group query message to all ports;

s53, after receiving the ordinary group inquiry message after framing, the user side device supporting IGMP protocol responds to generate a group adding message and sends the group adding message to all switches, wherein the group adding message comprises the IP multicast address of the user side device supporting IGMP protocol; the exchanger receives the grouping message and sends the grouping message to the CPU of the exchanger; the CPU of the CPU analyzes the grouping message, extracts the IP multicast address in the grouping message, and forms a mapping relation table of the IP multicast address in the grouping message and the port where the IP multicast address is located.

It should be noted that no matter which switch in the cluster is selected to issue the normal group query request, the normal group query message can reach all ports in the cluster, and because of the relationship of the aggregation group, the peer link will not transmit the normal group query message, and will not cause broadcast storm in the ring cluster.

For S6, specifically, when the switch in the ring cluster receives the IP multicast service frame, the mapping relation table is queried to find a destination port, and the IP multicast service frame is forwarded to the destination port, so as to implement a working process of forwarding the IP multicast service by the ring cluster, which may include:

s61, after receiving the IP multicast service frame, the exchanger in the ring cluster analyzes the IP multicast address in the IP multicast service frame;

s62, inquiring a mapping relation table according to the analyzed IP multicast address to obtain a corresponding destination port;

s63, the received IP multicast service frame is sent to a corresponding destination port; and forwarding the received IP multicast service frame to two adjacent switches through the inter-machine interconnection port.

In addition, it should be noted that the grouping message is only uploaded to the directly connected switch through the user terminal device, the switch receiving the grouping message sends the grouping message to the CPU, the CPU selectively sends the grouping message to the inter-machine interconnection port while storing the grouping message in the own hardware FPGA board, and finally all switches in the cluster receive the grouping message. The memory resources required for this entry are extremely high in practical applications, especially when the switch is a multi-bus processing architecture, embodiments of the present invention provide one such entry for each bus to ensure query efficiency. Because the storage resources of the switches realized by using the FPGA are extremely limited, the method provided by the embodiment of the invention can be used for setting the clusters, and can be used for sharing the table entries when the large-scale IP multicast service arrives, so that the storage pressure of each switch in the clusters can be greatly reduced.

The ring switch cluster has high reliability, and can complete forwarding through other paths after the link is damaged. The method can effectively solve the problems of side link faults, inter-machine link faults and network side link faults of users.

The embodiment of the invention provides a data flow forwarding schematic diagram of a user side link failure of a two-layer ring-shaped reliable switch clustering method of a large-scale IP multicast service, please refer to fig. 7. The user side link failure refers to that the user side device accesses two switches in the aggregation group by means of dual homing access, taking the user side device 1 in fig. 7 as an example, after the configuration is completed, the link accessing port3 of switch 1 and the link accessing port3 of switch 2 are equivalent for the user side device 1. The two links are mutually backup, and load balancing can be realized by the user terminal equipment 1. Blue arrow indicates that the user equipment 1 at the user side transmits a data stream to the user equipment 3 at the network side, and when the popr 3 link accessed to the switch 1 is damaged, the data can still enter the ring switch cluster through the backup link, namely the port3 link of the switch 2, and is forwarded to the user equipment 3 through the switches 2, 4 and 3. After arriving at switch 1, the data traffic arriving from the network side by the red arrow may now release the forwarding restriction of the peer link because of the link failure at port3 of switch 1, reach switch 2 of its same group via the peer link, and be sent to customer premises equipment 1 through port3 of switch 2. Such a design greatly improves the high reliability of the ring switch cluster.

The embodiment of the invention provides a data flow forwarding schematic diagram of inter-machine link failure of a two-layer ring-shaped reliable switch clustering method of a large-scale IP multicast service, please refer to fig. 8. The inter-machine link failure refers to that the inter-machine link of the ring network is damaged, so that transmission is interrupted, taking the link between the switch 1 and the switch 3 in fig. 8 as an example, the blue arrow is still that the user side device 1 sends a data stream to the network side user side device 3, and due to the ring cluster structure, the inter-machine link damage of one switch does not affect the forwarding of the data, and the data can bypass the switch 2-switch 4-switch 3 and be forwarded to the user side device 3. It should be noted that, the load balance exists in the ue 1, that is, different loads are transmitted on two equivalent aggregation links, when there is an inter-device link failure, the data cannot be directly forwarded to the switch 3 after reaching the switch 1, and at this time, the forwarding restriction of the peer_link may be released, and the data reaches the switch 2 in the same group with the peer_link via the peer_link and is forwarded to the ue 3 via the switch 2-switch 4-switch 3 in sequence. After arriving at switch 3, the data traffic arriving from the network side by the red arrow cannot pass directly to switch 1 via the shortest path, but can still bypass through switch 3-switch 4-switch 2 to customer premises equipment 1.

The embodiment of the invention provides a data flow forwarding schematic diagram of network side link failure of a two-layer ring-shaped reliable switch clustering method of a large-scale IP multicast service, please refer to fig. 9. The network side link failure refers to that the link between the ordinary switch used by the network side of the ring switch cluster to access the network side device and the network side device is damaged, taking the link between the port4 of the switch 3 and the network side device in fig. 9 as an example, it can be seen from the figure that the ring cluster has two ordinary switches, that is, two link access network side devices, that is, the link between the port4 of the switch 3 and the network side device, and the link between the port4 of the switch 4 and the network side device, respectively. The blue arrow is the data flow sent from the user side device 1 to the network side device, and after the data flow arrives at the switch 3, the network side link failure cannot enter the network side device, but still can pass through another link connected with the network side device, and is finally sent into the network side device through the switch 4. The same holds true for the data traffic coming from the network side with red arrows, and when one access network link fails, the data traffic can still be sent to the ring switch cluster via the other access network link, and finally reach the customer premise equipment 1 by forwarding.

In summary, the ring switch cluster provided by the embodiment of the invention is composed of 4 switches, and the two switches are located at the user side and are aggregated into an aggregation switch to provide reliability for access of user data; the rest two common switches are positioned at the network side, provide two links for accessing the network terminal equipment, provide reliability for the access of the network terminal equipment, and simultaneously can access the user terminal equipment with low requirements on reliability to enter the cluster; and the reliability of the inter-machine links in the annular cluster is ensured through the annular structure of 4 switches. The flexible coordination of three mechanisms realizes high system link reliability and bandwidth while ensuring the scale of the access equipment. Each switch in the cluster is mutually matched so as to complete the two-layer forwarding of the large-scale IP multicast service.

Compared with the prior art, the embodiment of the invention has the following advantages:

(1) The cluster structure formed by 4 switches is adopted, so that the access scale is larger, more end devices are supported compared with the common switches, and the switches can still be added into the cluster model for ring connection so as to enhance expansibility;

(2) Under the condition of limited FPGA resources, a larger forwarding table item can be maintained in the cluster, so that the forwarding capacity and the processing speed of the IP multicast service are improved;

(3) When one link of the user side equipment fails, the aggregation switch selects other links to forward, and two switches in the aggregation switch are redundant, so that the reliability and redundancy of the equipment are improved;

(4) By means of annular cluster design, the switch 3 and the switch 4 are respectively connected with the network side, and when one inter-machine link of the switch fails or the network side link fails, information forwarding is guaranteed, so that higher reliability and load balancing are achieved.

The foregoing description is only of the preferred embodiments of the present invention and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims (10)

1. A method for clustering two-layer ring-shaped reliable switches for large-scale IP multicast services, comprising:

interconnecting switch 1, switch 2, switch 3 and switch 4 into 1 ring network, thereby forming a ring cluster;

configuring the adjacent switch 1 and the adjacent switch 2 as an aggregation switch; the aggregation switch is used for accessing user equipment and special requirement equipment;

setting the switch 3 and the switch 4 as a normal switch; the switch 3 and the switch 4 are both used for accessing network end equipment and user end equipment;

dividing the ports of the aggregation switch, the switch 3 and the switch 4 into inter-machine interconnection ports and common end equipment access ports;

adopting modified IGMP snoring configuration to the annular cluster to obtain a mapping relation table of an IP multicast service frame and a destination port;

when receiving the IP multicast service frame, the switch in the ring cluster finds a destination port corresponding to the IP multicast service frame received by the switch according to the mapping relation table, and sends the IP multicast service frame to the corresponding destination port, and meanwhile forwards the IP multicast service frame to an adjacent switch, so that the two-layer forwarding of the IP multicast service is realized.

2. A two-layer ring reliable switch clustering method for large scale IP multicast traffic according to claim 1, wherein the CPU of said switch 1 and said switch 2 are configured with aggregate group information; the aggregation group information includes: group number and priority information; the group number is an identity in an aggregation group formed by the switch 1 and the switch 2; the priority information indicates a priority level of the switch operation.

3. A method for two-layer ring reliable switch clustering for large scale IP multicast traffic according to claim 2, wherein said configuring adjacent said switch 1 and said switch 2 as an aggregate switch comprises:

configuring the aggregation group information by the CPU of the switch 1 and the CPU of the switch 2, and configuring the aggregation group information to respective hardware FPGA boards through PIO;

the hardware FPGA board cards of the switch 1 and the switch 2 carry out framing encapsulation on the received aggregation group information after the configuration respectively, and broadcast a local port as an aggregation group hello message so as to send the aggregation group hello message to an adjacent switch;

after the adjacent exchanger receives the corresponding aggregation group hello message, inquiring whether the adjacent exchanger meets the preset condition is as follows: the group number of the aggregation group is the same as the group number of the aggregation group in the received aggregation group hello message; if yes, determining a group of switches for sending and receiving the aggregation group hello message as the switches to be paired; the ports of the switches to be paired, which are respectively responsible for sending and receiving the aggregation group hello messages, are connected with each other and are determined to be the interconnection ports in the groups, so that an aggregation switch is formed;

and when the aggregation is completed, the switch receiving the corresponding aggregation group hello message compares the priority information in the received aggregation group hello message with the own priority information, and confirms the master device and the slave device.

4. A method for a two-layer ring reliable switch cluster for large-scale IP multicast traffic as claimed in claim 3, wherein the switch receiving said corresponding aggregate hello message compares priority information in said corresponding aggregate hello message with self priority information, and validates master and slave devices, comprising:

the exchanger receiving the aggregation group hello message judges whether the priority information in the received aggregation group hello message is greater than the priority information of the exchanger per se;

if yes, determining a switch for sending the aggregation group hello message as a master device, and determining a switch for receiving the aggregation group hello message as a slave device;

if not, determining the exchanger for receiving the aggregation group hello message as a master device, and determining the exchanger for sending the aggregation group hello message as a slave device.

5. The method for two-layer ring reliable switch clustering of large-scale IP multicast traffic of claim 4 wherein said intra-group interconnect port links are peer link links.

6. The method for two-layer ring reliable switch clustering of large-scale IP multicast traffic according to claim 5, wherein the dividing the ports of the aggregation switch, the switch 3 and the switch 4 into the division pattern of the inter-machine interconnect ports and the access ports of the normal end devices comprises:

manual division mode and automatic recognition division mode.

7. The method for two-layer ring reliable switch clustering for large-scale IP multicast traffic of claim 6, wherein said manual partitioning mode of operation comprises:

after receiving a setting instruction of a user, a CPU of the current switch transmits an inter-machine interconnection port number of the current switch to a hardware FPGA board card of the current switch, and the hardware FPGA board card of the current switch is maintained.

8. The method for clustering two-layer ring-type reliable switches for large-scale IP multicast service according to claim 7, wherein said automatically identifying the operation of the partitioning mode comprises:

the CPU of the current switch issues a port discovery instruction to a hardware FPGA board card of the current switch, the hardware FPGA board card of the current switch carries out framing after perceiving the port discovery instruction, the information after framing comprises a switch number of the current switch, and the current switch forwards the information after framing in a broadcasting mode;

when the port of the exchanger receives the information carrying the framing of the non-self exchanger number, the port receiving the information after framing is set as an inter-machine interconnection port, and the other ports are set as common terminal equipment access ports;

when the network access environment changes, the port division is carried out again.

9. The method for two-layer ring reliable switch clustering of large-scale IP multicast service according to claim 8, wherein the working process of adopting modified IGMP snoring configuration to the ring cluster to obtain the mapping relation table of the IP multicast service frame and the destination port comprises the following steps:

1 exchanger is selected in the ring cluster, and a common group query request is issued through the CPU of the selected exchanger;

after receiving the normal group query request, the hardware FPGA board card of the selected switch carries out framing of a normal group query message, and sends the framed normal group query message to all ports;

after receiving the ordinary group query message after framing, the user side equipment supporting the IGMP protocol responds to generate a group adding message and sends the group adding message to a switch connected with the user side equipment, wherein the group adding message comprises the IP multicast address of the user side equipment supporting the IGMP protocol; the exchanger connected with the user side equipment receives the grouping message and sends the grouping message to the CPU of the exchanger; and the CPU of the CPU analyzes the grouping message, extracts the IP multicast address in the grouping message, and forms a mapping relation table of the IP multicast address in the grouping message and the port where the IP multicast address is located.

10. The method for a two-layer ring reliable switch cluster of a large-scale IP multicast service according to claim 9, wherein, after the switch in the ring cluster receives an IP multicast service frame, the switch searches the mapping relation table to find a destination port, and forwards the IP multicast service frame to the destination port, so as to implement a working process of forwarding the IP multicast service by the ring cluster, which includes:

after receiving the IP multicast service frame, the exchanger in the ring cluster analyzes the IP multicast address in the IP multicast service frame, queries the mapping relation table according to the analyzed IP multicast address to obtain a corresponding destination port, and sends the received IP multicast service frame to the corresponding destination port; and forwarding the received IP multicast service frame to two adjacent switches through the inter-machine interconnection port.