Entry synchronization method and apparatus, network device, and storage medium

Abstract

An entry synchronization method and apparatus, a network device, and a storage medium, relating to the technical field of communications. The method is applied to a first forwarding node, the first forwarding node is located in an AI network, and the first forwarding node is connected to a host by means of a first port. The method comprises: receiving a first packet sent by a host; on the basis of the first packet, learning a first global scheduling entry corresponding to the host, wherein the first global scheduling entry comprises a system port identifier of the first port, and the system port identifier comprises a device identifier and a port identifier; and sending a second packet to a second forwarding node, wherein the second packet comprises the first global scheduling entry, so that the second forwarding node learns a second global scheduling entry, wherein the second global scheduling entry is the same as the first global scheduling entry. The solution can reduce the impact of entry synchronization on a control plane, so that the performance and expansion capability of the control plane is improved.

Description

Table entry synchronization method and device, network equipment and storage medium TECHNICAL FIELD

[0001] The present application relates to the technical field of communication, in particular to a table entry synchronization method and device, network equipment and storage medium. BACKGROUND

[0002] An artificial intelligence (AI) network has the characteristics of few flows and large flows. In order to ensure the load balancing of the AI network, the full scheduling Ethernet technology is used in the AI network to perform load sharing on a packet-by-packet or cell-by-cell basis. Specifically, the full scheduling Ethernet technology can be as follows:

[0003] The first node encapsulates scheduling information for the cells of each packet according to a full scheduling table entry; and the multiple cells are transmitted to a tail node in a load-balanced manner through multiple forwarding paths between the first node and the tail node. The tail node assembles the multiple cells into a complete packet according to the full scheduling table entry and the scheduling information carried by the cells of each packet, and performs out-of-order rearrangement on the multiple packets.

[0004] In the above full scheduling Ethernet technology, the full scheduling table entry needs to be synchronized between the first node and the tail node. However, the synchronization of the full scheduling table entry is currently implemented by a controller of a control plane, which seriously affects the performance and expansion capability of the control plane.

[0005] SUMMARY

[0006] Embodiments of the present application aim to provide a table entry synchronization method and device, network equipment and storage medium to reduce the impact of table entry synchronization on the control plane and improve the performance and expansion capability of the control plane. The specific technical solutions are as follows:

[0007] In a first aspect, the embodiments of the present application provide a table entry synchronization method applied to a first forwarding node, wherein the first forwarding node is in an artificial intelligence (AI) network, and the first forwarding node is connected to a host through a first port. The method comprises the following steps:

[0008] receiving a first packet sent by the host;

[0009] learning a first full scheduling table entry corresponding to the host according to the first packet, wherein the first full scheduling table entry comprises a system port identifier of the first port, and the system port identifier comprises a device identifier and a port identifier;

[0010] sending a second packet to a second forwarding node in the AI network, wherein the second packet comprises the first full scheduling table entry, so that the second forwarding node learns a second full scheduling table entry, and the second full scheduling table entry is the same as the first full scheduling table entry.

[0011] In some embodiments, the learning, according to the first packet, of the first full scheduling table item corresponding to the host specifically comprises:

[0012] The source address included in the first packet is acquired, and the system port identifier of the first port is acquired;

[0013] The first full scheduling table item corresponding to the host is learned according to the acquired source address and system port identifier.

[0014] In some embodiments, the second packet includes a tunnel encapsulation Type Length Value (TLV), and the tunnel encapsulation TLV includes a type field, a length field, and a value field;

[0015] The type field is used to indicate that the value field carries tunnel encapsulation information; the length field is used to carry a length value of the tunnel encapsulation information; and the value field is used to carry the tunnel encapsulation information.

[0016] The value field includes a type subfield, a length subfield, and a value subfield; the type subfield is used to indicate that the value subfield carries one system port identifier; the length subfield is used to carry a length value of the one system port identifier; and the value subfield is used to carry the one system port identifier.

[0017] In some embodiments, the second packet is an Ethernet Virtual Private Network (EVPN) class II route packet, and a network reachable information label 2 field included in the EVPN class II route packet carries the system port identifier.

[0018] In some embodiments, the second packet includes an EVPN class II route packet and an EVPN class V route packet;

[0019] The network reachable information label 2 field included in the EVPN class II route packet carries the system port identifier.

[0020] The gateway address field included in the EVPN class V route packet carries an Internet Protocol (IP) address of the host.

[0021] In some embodiments, the step of learning, according to the first packet, of the first full scheduling table item corresponding to the host comprises:

[0022] The system port identifier of the first port used to receive the first packet is acquired;

[0023] add the system port identifier to a multicast table corresponding to a Virtual Switching Instance (VSI) to which the host belongs, to obtain a first full scheduling table item corresponding to the host.

[0024] In some embodiments, the second packet includes a tunnel encapsulation TLV, the tunnel encapsulation TLV includes a type field, a length field and a value field;

[0025] The type field is used to indicate that the value field carries tunnel encapsulation information; the length field is used to carry a length value of the tunnel encapsulation information; and the value field is used to carry the tunnel encapsulation information.

[0026] The value field includes a type subfield, a length subfield and a value subfield; the type subfield is used to indicate that the value subfield carries a plurality of system port identifiers; the length subfield is used to carry a length value of the plurality of system port identifiers; and the value subfield is used to carry the plurality of system port identifiers.

[0027] The value subfield includes a length sub-subfield and at least one value sub-subfield; the length sub-subfield is used to indicate a number of the plurality of system port identifiers; and the value sub-subfield is used to carry one system port identifier.

[0028] In some embodiments, the type subfield is further used to indicate that the value subfield carries an encapsulation index; the length subfield is further used to carry a length value of the encapsulation index; and the value subfield is further used to carry the encapsulation index.

[0029] In a second aspect, embodiments of the present application provide a table item synchronization device, applied to a first forwarding node, the first forwarding node being in an artificial intelligence (AI) network, the first forwarding node being connected to a host through a first port, and the device comprising:

[0030] a receiving module, configured to receive a first packet sent by the host;

[0031] a learning module, configured to learn, according to the first packet, a first full scheduling table item corresponding to the host, the first full scheduling table item including a system port identifier of the first port, and the system port identifier including a device identifier and a port identifier;

[0032] a synchronization module, configured to send a second packet to a second forwarding node in the AI network, the second packet including the first full scheduling table item, so that the second forwarding node learns a second full scheduling table item, the second full scheduling table item being the same as the first full scheduling table item.

[0033] In some embodiments, the learning module is specifically configured to: acquire a source address included in the first packet, and acquire a system port identifier of the first port; and learn a first full scheduling table item corresponding to the host according to the acquired source address and system port identifier.

[0034] In some embodiments, the second packet includes a tunnel encapsulation TLV, and the tunnel encapsulation TLV includes a type field, a length field and a value field.

[0035] The type field is used to indicate that the value field carries tunnel encapsulation information; the length field is used to carry a length value of the tunnel encapsulation information; and the value field is used to carry the tunnel encapsulation information.

[0036] The value field includes a type subfield, a length subfield and a value subfield; the type subfield is used to indicate that the value subfield carries a system port identifier; the length subfield is used to carry a length value of the system port identifier; and the value subfield is used to carry the system port identifier.

[0037] In some embodiments, the second packet is an EVPN class II route packet, and a network reachable information label 2 field included in the EVPN class II route packet carries the system port identifier.

[0038] In some embodiments, the second packet includes an EVPN class II route packet and an EVPN class V route packet.

[0039] The network reachable information label 2 field included in the EVPN class II route packet carries the system port identifier.

[0040] The gateway address field included in the EVPN class V route packet carries an address of the host in the EVPN class II route.

[0041] In some embodiments, the learning module is specifically configured to: acquire a system port identifier of the first port, the first port being used to receive the first packet; and add the system port identifier to a multicast table corresponding to a virtual switching instance to which the host belongs, to obtain a first full scheduling table item.

[0042] In some embodiments, the second packet includes a tunnel encapsulation TLV, and the tunnel encapsulation TLV includes a type field, a length field and a value field.

[0043] The type field is used to indicate that the value field carries tunnel encapsulation information; the length field is used to carry a length value of the tunnel encapsulation information; and the value field is used to carry the tunnel encapsulation information.

[0044] The value field comprises a type subfield, a length subfield and a value subfield; the type subfield is used to indicate that the value subfield carries a plurality of system port identifiers; the length subfield is used to carry a length value of the plurality of system port identifiers; and the value subfield is used to carry the plurality of system port identifiers.

[0045] The value subfield comprises a length sub-subfield and at least one value sub-subfield; the length sub-subfield is used to indicate a number of the plurality of system port identifiers; and the value sub-subfield is used to carry one system port identifier.

[0046] In some embodiments, the type subfield is further used to indicate that the value subfield carries an encapsulation index; the length subfield is further used to carry a length value of the encapsulation index; and the value subfield is further used to carry the encapsulation index.

[0047] In a third aspect, an embodiment of the present application provides a network device, comprising a processor and a machine readable storage medium, the machine readable storage medium stores a computer program capable of being executed by the processor, and the processor is prompted by the computer program to implement any of the method steps described above.

[0048] In a fourth aspect, an embodiment of the present application provides a machine readable storage medium, the machine readable storage medium stores a computer program, and the computer program is executed by a processor to implement any of the method steps described above.

[0049] In a fifth aspect, an embodiment of the present application provides a computer program, and the computer program is executed by a processor to implement any of the method steps described above.

[0050] In the technical solution provided by the embodiments of the present application, the forwarding node carries the full scheduling table item in a message and sends it to other forwarding nodes, so that other forwarding nodes learn the same full scheduling table item, i.e., the full scheduling table item includes the system port identifier of the connected port when the host accesses the forwarding node. In the embodiments of the present application, the synchronization of the full scheduling table item is achieved by transmitting messages between the forwarding nodes, and the controller is not needed to synchronize the full scheduling table item, thereby reducing the influence of the synchronization of the table item on the control plane and improving the performance and expansion capability of the control plane. BRIEF DESCRIPTION OF DRAWINGS

[0051] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and serve to explain the principles of the application, and do not limit the application in any way.

[0052] FIG. 1 is a first schematic diagram of an AI network architecture;

[0053] FIG. 2 is a second schematic diagram of an AI network architecture;

[0054] Fig. 3 is a third schematic diagram of the AI network architecture;

[0055] Fig. 4 is a flow diagram of a table entry synchronization method provided by an embodiment of the present application;

[0056] Fig. 5 is a first schematic diagram of step S402 provided by an embodiment of the present application;

[0057] Fig. 6a is a schematic diagram of an extended TLV field provided by an embodiment of the present application;

[0058] Fig. 6b is a first schematic diagram of a value field in the extended TLV field shown in Fig. 6a;

[0059] Fig. 6c is a second schematic diagram of the value field in the extended TLV field shown in Fig. 6a;

[0060] Fig. 6d is a third schematic diagram of the value field in the extended TLV field shown in Fig. 6a;

[0061] Fig. 7 is a schematic diagram of an EVPN class II route provided by an embodiment of the present application;

[0062] Fig. 8 is a schematic diagram of an EVPN class V route provided by an embodiment of the present application;

[0063] Fig. 9 is a second schematic diagram of step S402 provided by an embodiment of the present application;

[0064] Fig. 10 is a schematic diagram of the AI network architecture provided by an embodiment of the present application;

[0065] Fig. 11 is a schematic diagram of an internal processing flow of a device provided by an embodiment of the present application;

[0066] Fig. 12 is a schematic diagram of a structure of a table entry synchronization apparatus provided by an embodiment of the present application;

[0067] Fig. 13 is a schematic diagram of a structure of a network device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0068] To make the objectives, technical solutions, and advantages of the present application clearer, further detailed descriptions will be made to the present application with reference to the accompanying drawings and embodiments. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art belong to the scope of protection of the present application.

[0069] With the development of communication technology, the wide application of AI network technology has become an important driving force in various industries. AI network adopts Spine-Leaf architecture, as shown in FIG. 1. The Spine-Leaf architecture is a network topology composed of two switching layers of Spine and Leaf. The Leaf switching layer contains a two-layer switch (i.e., Leaf switch) for connecting servers and storage devices, and the Leaf switch is used to aggregate traffic from servers and storage devices; the Spine switching layer contains a three-layer switch (i.e., Spine switch) for routing and forwarding, which belongs to the backbone of the network.

[0070] In the Spine-Leaf architecture, each Leaf switch is connected to a Spine switch. Through this design, each server in the network only needs to pass through the Leaf switch and the Spine switch when transmitting data with other servers, greatly improving the efficiency of data transmission. Therefore, the Spine-Leaf architecture can provide high-bandwidth, low-latency, non-blocking server-to-server connections, and can realize efficient scheduling and management of GPU resources to support the needs of large-scale AI workloads. This Spine-Leaf architecture is particularly important in high-performance computing cluster applications.

[0071] The Spine-Leaf architecture is also known as a distributed core network architecture. Since this network architecture is derived from the switch matrix module (Switch Fabric) inside the switch, the Spine-Leaf architecture is also known as the Fabric network architecture, and belongs to the CLOS network model.

[0072] In an interconnected network, forwarding nodes (such as the above-mentioned Leaf switch and Spine switch) perform load balancing forwarding of data flows based on N-tuple information of the data flow. Specifically, it can be: calculating the hash value of the N-tuple information of the data flow, determining the forwarding path corresponding to the hash value, and the packets of the data flow are all forwarded along the forwarding path.

[0073] This load balancing algorithm does not consider the flow size of the data flow, but only considers the N-tuple information of the data flow. In a multi-flow large-flow network (i.e., a network with a large number of data flows and small data flow traffic), using the above load balancing algorithm to forward data flows can effectively ensure load balancing and improve network throughput.

[0074] However, the AI network has the characteristics of few flows and large flows (i.e., few data flows and large data flow traffic). When the forwarding node adopts the above load balancing algorithm to forward the data flow in the AI network, the data flow is likely to be allocated to a few forwarding paths. As shown in the AI network in FIG. 2, which includes four Leaf switches Leaf1-Leaf4 and two Spine switches Spine1-Spine2, Leaf1-Leaf4 are connected to servers A-H. When server A sends a data flow to server H, Leaf1 adopts the above load balancing algorithm and is likely to allocate the data flow sent by server A to server H to the thick solid line paths shown in FIG. 2, i.e., the packets of the data flow sent by server A to server H are all forwarded along the thick solid line paths shown in FIG. 2. This causes load imbalance.

[0075] To solve the above problem of load imbalance, in the related art, full scheduling Ethernet technology is used to perform load sharing on a packet-by-packet or cell-by-cell basis, as follows:

[0076] (1) Packet-by-packet transmission.

[0077] Each Leaf switch stores a full scheduling table item, and the full scheduling table item is a forwarding table item supporting full scheduling Ethernet technology. The full scheduling table item can include an Internet Protocol (IP) address, a Media Access Control (MAC) address, a System Port Identity (SysPort ID), an encapsulation ID (i.e., an encapsulation index), and the like. The SysPort ID can uniquely identify a port connected to a host on the Leaf switch in the AI network. The SysPort ID includes a device ID and a port ID.

[0078] After the head node (Leaf switch) receives a packet from a source host, the head node encapsulates scheduling information for the packet according to the full scheduling table item. The scheduling information includes a SysPort ID, a sequence number (Sequence) of the packet, an IP address, a MAC address, or an encapsulation ID. The head node forwards the packet to a tail node (Leaf switch) through multiple forwarding paths between the head node and the tail node in a polling manner.

[0079] The tail node receives multiple packets sent by the head node through multiple forwarding paths, determines multiple packets belonging to the same data flow based on the scheduling information carried by the packets according to the full scheduling table item, and performs out-of-order rearrangement on the multiple packets. The tail node sends the rearranged packets to a destination host.

[0080] (2) Cell-by-cell transmission.

[0081] Each Leaf switch stores a full scheduling table entry, which is a forwarding table entry that supports full scheduling Ethernet technology. This full scheduling table entry may include IP address, MAC address, SysPort ID, encapsulation ID, etc.

[0082] After receiving a message from the source host, the first node segments the message into multiple cells; according to the full scheduling table entry, it encapsulates scheduling information for each cell, which includes SysPort ID, the message's sequence number, IP address, MAC address, or encapsulation ID; and forwards the cells to the tail node in a round-robin manner through multiple forwarding paths between the first node and the tail node (Leaf switch).

[0083] The tail node receives multiple cells sent by the head node through multiple forwarding paths. According to the full scheduling table entries and based on the scheduling information carried by the cells, it determines multiple cells belonging to the same data stream and the same message, assembles the multiple cells into a complete message, and rearranges the multiple messages in out of order; then it sends the rearranged message to the destination host.

[0084] By employing fully scheduled Ethernet technology, an authorized trust domain can be deployed within the AI ​​network without the need to deploy Explicit Congestion Notification (ECN) / Priority-based Flow Control (PFC) domains, as shown in Figure 3. This implements a mechanism of scheduling before transmission, achieving preventative congestion control and deterministic precise congestion control, thus achieving an optimal balance between congestion and link load.

[0085] In AI networks, whether transmitting packets or cells, the first and last nodes need to synchronize all scheduling entries. Currently, this synchronization is achieved by the controller shown in Figure 2, which severely impacts the performance and scalability of the control plane.

[0086] To address the aforementioned issues, this application provides an entry synchronization method, as shown in Figure 4, applied to a first forwarding node. The first forwarding node is located in an AI network and connects to a host via a first port. The method includes the following steps:

[0087] Step S401: Receive the first message sent by the host;

[0088] Step S402: Based on the first message, learn the first full scheduling table entry corresponding to the host. The first full scheduling table entry includes the system port identifier of the first port. The system port identifier includes the device identifier and the port identifier.

[0089] Step S403: sending a second packet to a second forwarding node in the AI network, the second packet comprising the first full scheduling table item, so that the second forwarding node learns the second full scheduling table item, and the second full scheduling table item is the same as the first full scheduling table item.

[0090] In the technical solution provided in the embodiments of the present application, the forwarding node carries the full scheduling table item in the packet and sends it to other forwarding nodes, so that other forwarding nodes learn the same full scheduling table item, i.e., the system port identifier of the connected port when the host accesses the forwarding node is included in the full scheduling table item. In the embodiments of the present application, the synchronization of the full scheduling table item is realized by transmitting the packet between the forwarding nodes, and the controller is not needed to synchronize the full scheduling table item, thereby reducing the influence of the synchronization of the table item on the control plane and improving the performance and expansion capability of the control plane.

[0091] In the embodiments of the present application, the first forwarding node and the second forwarding node can be any Leaf switch in the AI network. The first forwarding node and the second forwarding node can also be referred to as a Network Cloud Packet-Forwarder (NCP) node. The first port can be any port on the first forwarding node, and the host accesses the first forwarding node through the first port.

[0092] In the above step S401, the first packet can be a packet actively sent by the host, such as a packet sent by the host to other hosts. The first packet can also be a response packet fed back according to the packet sent by the first forwarding node, which is not limited.

[0093] After the host accesses the first forwarding node, the host can send a packet, i.e., the first packet, to the first forwarding node, and the first forwarding node receives the first packet through the first port.

[0094] In the above step S402, the first full scheduling table item is a full scheduling table item corresponding to the host. The full scheduling table item can be a two-layer table item, such as a MAC table item, at this time, the full scheduling table item can include the MAC address of the host, the SysPort ID, etc., such as the MAC table item shown in Table 1.

[0095] Table 1

[0096] The full scheduling table item can also be a three-layer table item, such as an Address Resolution Protocol (ARP) table item and a Forwarding Information Base (FIB) table item, at this time, the full scheduling table item can include the IP address of the host, the MAC address of the host, the next hop IP address, the SysPort ID, etc., such as the ARP table item shown in Table 2 and the FIB table item shown in Table 3.

[0097] Table 2

[0098] Table 3

[0099] The first forwarding node learns the full scheduling table item in an active or passive manner, as follows:

[0100] (1) The first forwarding node learns the full scheduling table item in an active manner.

[0101] After detecting that the host accesses the first forwarding node, the first forwarding node sends an ARP request to the host. Then, the host feeds back an ARP response (e.g., the first message) corresponding to the ARP request to the first forwarding node. The first forwarding node extracts the required information from the ARP response to generate the full scheduling table item (e.g., the first full scheduling table item) corresponding to the host.

[0102] (2) The first forwarding node learns the full scheduling table item in a passive manner.

[0103] After the host accesses the first forwarding node, the first forwarding node does not learn the full scheduling table item corresponding to the host. When the message (e.g., the first message) sent by the host reaches the first forwarding node, the first forwarding node extracts the required information from the message to generate the full scheduling table item corresponding to the host.

[0104] In the embodiments of the present application, the manner in which the first forwarding node learns the full scheduling table item is not limited.

[0105] In the above step S403, after learning the first full scheduling table item, the first forwarding node generates a second message and sends the second message to the second forwarding node, where the second message includes the information of the first full scheduling table item. After receiving the second message, the second forwarding node can extract the information of the first full scheduling table item from the second message, learn the second full scheduling table item, and the second full scheduling table item is the same as the first full scheduling table item, that is, the second full scheduling table item includes the system port identifier of the first port.

[0106] In the embodiments of the present application, the first full scheduling table item can be a table item for implementing unicast or a table item for implementing multicast, and the multicast can include broadcast and groupcast.

[0107] When the first full scheduling table item is a table item for implementing unicast, as shown in FIG. 5, the above step S402 can be as follows:

[0108] In step S501, the source address included in the first message is obtained, and the system port identifier of the first port is obtained.

[0109] In step S502, the first full scheduling table item corresponding to the host is learned according to the obtained source address and system port identifier.

[0110] In the embodiments of the present application, the source address of the first packet is the address of the host. The first forwarding node obtains the required information according to the established table item type, such as the MAC address and SysPort ID of the host when the second layer table item is established, the IP address, the MAC address and SysPort ID of the host when the third layer table item is established, and then establishes the corresponding full scheduling table item according to the obtained information.

[0111] In some embodiments, to realize the synchronization of the full scheduling table item shown in FIG. 5, the second packet includes a tunnel encapsulation TLV, which is an added extended TLV. The tunnel encapsulation TLV can include a type field, a length field and a value field, as shown in FIG. 6a.

[0112] The type field is used to indicate that the value field carries tunnel encapsulation information, such as the type 1 field shown in FIG. 6a, which has a length of 2 octets; the length field is used to carry the length value of the tunnel encapsulation information, that is, the length field is used to carry the length value of the value field, such as the length 1 field shown in FIG. 6a, which has a length of 2 octets; and the value field is used to carry the tunnel encapsulation information, such as the value 1 field shown in FIG. 6a, which has a variable length.

[0113] In the embodiments of the present application, as shown in FIG. 6b, the value field can include a type subfield, a length subfield and a value subfield. The type subfield is used to indicate that the value subfield carries a system port identifier, such as the type 2 field shown in FIG. 6b, which has a length of 1 octet; the length subfield is used to carry the length value of the system port identifier, that is, the length subfield is used to carry the length value of the value subfield, such as the length 2 field shown in FIG. 6b, which has a length of 1 octet; and the value subfield is used to carry the system port identifier, such as the value 2 field shown in FIG. 6b, which has a length of 2 octets.

[0114] In the embodiments of the present application, the lengths of the various fields shown in FIGS. 6a and 6b are only examples and are not limiting. The first forwarding node can use the Border Gateway Protocol (BGP) extended route attribute to carry the SysPort ID in the second packet.

[0115] In the embodiments of the present application, to facilitate implementation and reduce adjustment to the existing protocol, the first forwarding node can use the existing route packet to carry the SysPort ID.

[0116] In an embodiment, the second message comprises an EVPN type-2 route message, and the NLRI label 2 field of the EVPN type-2 route message carries the SysPort ID.

[0117] As shown in FIG. 7, the EVPN type-2 route message comprises a Route Distinguisher (RD) field, an Ethernet Segment Identifier (ESI) field, an Ethernet Tag ID field, a MAC Address Length field, a MAC Address field, an IP Address Length field, an IP Address field, a Multi-Protocol Label Switching (MPLS) label 1 field, and an MPLS label 2 field. The MPLS label 2 field is the NLRI label 2 field described above. In an embodiment of the present application, the first forwarding node can fill the SysPort ID in the high bits of the NLRI label 2 field, such as 2 bytes of high bits, to fill the SysPort ID.

[0118] With the EVPN type-2 route message, the first forwarding node can send the second forwarding node the full scheduling table item of the Layer 2, such as the MAC table item.

[0119] In some embodiments, in order to implement the full scheduling table item of the Layer 3, the first forwarding node can synchronize the table item information of the full scheduling table item to the second forwarding node by using the EVPN type-2 route message described above, and also synchronize the table item information of the full scheduling table item to the second forwarding node by using the EVPN type-5 route message described above, so that the second forwarding node can learn the complete full scheduling table item.

[0120] That is, the second message described above can comprise the EVPN type-2 route message and the EVPN type-5 route message. The NLRI label 2 field of the EVPN type-2 route message carries the SysPort ID, which can be referred to the related description of FIG. 7. The gateway address field of the EVPN type-5 route message carries the IP address of the host. The IP address of the host is also carried in the IP address field of the EVPN type-2 route message, that is, the next hop address carried in the IP address field of the EVPN type-2 route message is the same as the address carried in the gateway address field.

[0121] The EVPN five-type route packet shown in FIG. 8 includes an RD field, an ESI field, an Ethernet label identification field, an IP prefix Length field, an IP prefix field, a gateway address (GW IP Address) field and an MPLS label field.

[0122] Based on the IP address carried in the gateway address field in the EVPN five-type route packet, the second forwarding node can find the corresponding EVPN two-type route packet, and further obtain all information required for learning the full scheduling table item, and learn the first full scheduling table item.

[0123] When the first full scheduling table item is a table item for implementing multicast, as shown in FIG. 9, the above step S402 can be:

[0124] In step S901, a system port identifier of a first port is obtained, the first port being used for receiving a first packet.

[0125] In step S902, the system port identifier is added to a multicast table corresponding to a VSI to which the host belongs, to obtain a first full scheduling table item corresponding to the host.

[0126] In the multicast scenario, in the embodiment, the first forwarding node obtains a system port identifier of a first port, and determines a VSI to which a host belongs, and adds the obtained system port identifier to a multicast table corresponding to the VSI. In the multicast table, a full scheduling table item corresponding to one host includes one system port identifier. After the multicast table is established, the first forwarding node can synchronize the multicast table to the second forwarding node through the BGP EVPN three-type route packet, so that the second forwarding node learns the same multicast table.

[0127] In some embodiments, to implement synchronization of the full scheduling table item shown in FIG. 9, the second packet includes a tunnel encapsulation TLV, the tunnel encapsulation TLV including an extended TLV added, the tunnel encapsulation TLV including a type field, a length field and a value field, as shown in FIG. 6a. The type field is used to indicate that the value field carries tunnel encapsulation information; the length field is used to carry a length value of the tunnel encapsulation information; and the value field is used to carry the tunnel encapsulation information.

[0128] In the embodiment, as shown in FIG. 6c, the value field includes a type subfield, a length subfield and a value subfield.

[0129] The type subfield is used to indicate that the value subfield carries a plurality of system port identifications, as shown in the type 3 field in FIG. 6c, the length of the type 3 field being 1 byte; the length subfield is used to carry the length value of the plurality of system port identifications, that is, the length subfield is used to carry the length value of the value subfield, as shown in the length 3 field in FIG. 6c, the length of the length 3 field being 2 bytes; and the value subfield is used to carry the plurality of system port identifications, the length of the value subfield being variable.

[0130] The value subfield includes a length sub-subfield and at least one value sub-subfield; the length sub-subfield is used to indicate the number of the plurality of system port identifications (the number of the value sub-subfields), as shown in the length 4 field in FIG. 6c, the length of the length 4 field being 1 byte; and the value sub-subfield is used to carry one system port identification, as shown in the plurality of value 3 fields in FIG. 6c, the length of the value 3 field being 2 bytes.

[0131] In the embodiments of the present application, the first forwarding node carries the SysPort ID by extending the routing message, to realize synchronization of the full scheduling table item. Based on the full scheduling table item, the second forwarding node can send the message to the first forwarding node in a packet-by-packet or cell-by-cell manner, to facilitate implementation of the message ordering function between the first forwarding node and the second forwarding node.

[0132] In some embodiments, to improve the forwarding efficiency of the data plane, the type subfield can also be used to indicate that the value subfield carries an encapsulation index (i.e., an encapsulation ID); and the length subfield is also used to carry the length value of the encapsulation index. In this case, the value subfield can also be used to carry the encapsulation index.

[0133] When the encapsulation ID is carried by the attribute of the extended routing message, the structures of the type subfield, the length subfield and the value subfield are as shown in FIG. 6d. The type 5 field shown in FIG. 6d is the type subfield, indicating that the value subfield carries the encapsulation index, the length of the type 5 field being 1 byte; the length 5 field shown in FIG. 6d is the length subfield, indicating the length value of the value subfield, the length of the length 5 field being 1 byte; and the value 5 field shown in FIG. 6d is the value subfield, carrying the encapsulation index of the type indicated by the type subfield, the length of the value 5 field being 2 bytes.

[0134] In the case where the encapsulation index is synchronized between the first forwarding node and the second forwarding node, when the second forwarding node sends the message or the cell to the first forwarding node, the scheduling information encapsulated by the message or the cell can carry the encapsulation index, and the IP address and the MAC address are omitted. The data amount of the encapsulation index is smaller than that of the IP address and the MAC address, and thus the second forwarding node sends the message or the cell to the first forwarding node, reduces the data amount of the transmitted data, and improves the forwarding efficiency of the data plane.

[0135] The table entry synchronization method provided by the embodiment of the present application is described in detail below in combination with the network architecture shown in FIG. 10. The network architecture shown in FIG. 10 includes two network cloud fabric (NCF) nodes, namely NCF1-NCF2, which can be understood as Spine switches. The network architecture shown in FIG. 10 also includes four NCP nodes, namely NCP1-NCP4, which can be understood as Leaf switches. The NCP1-NCP4 are connected to five servers, namely server1-server5. Take the synchronization of the full scheduling table entries from NCP4 to NCP1-NCP3 as an example for description. The NCP1-NCP4 are BGP peers.

[0136] (1) MAC table entry learning and synchronization.

[0137] The server4 sends a packet1 to the NCP4, and the source MAC address of the packet1 is the MAC address of the server4, such as aaa-bbb-ccc. After receiving the packet1, the NCP4 obtains the MAC address of the server4 and obtains the SysPort ID of port1 connected to the server4, such as drv1 port1. Wherein, drv1 is the device ID, and port1 is the port ID. Similarly, the NCP4 obtains the MAC address aaa-bbb-ddd of the server5 and obtains the SysPort ID of port2 connected to the server5, such as drv2 port1.

[0138] At this time, the NCP4 learns the MAC table entries corresponding to the server4 and the server5 as shown in Table 4.

[0139] Table 4

[0140] The NCP4 can use the fields shown in FIG. 6a and FIG. 6b to extend the BGP EVPN class-2 route packet, or use the multiplexing MPLS label2 field shown in FIG. 7 to carry the SysPort ID; and then synchronize the MAC table entries shown in Table 4 to the NCP1-NCP3 through the BGP EVPN class-2 route packet, so that the NCP1-NCP3 obtain and record the MAC table entries shown in Table 4.

[0141] After synchronizing the above MAC table entries, take the server1 sending a packet to the server4 as an example. The server1 sends a packet2, and the destination MAC address of the packet2 is the MAC address aaa-bbb-ccc of the server4. After receiving the packet2, the NCP1 looks up Table 4 according to the destination MAC address of the packet2, and obtains the out-interface as drv1 port1. The NCP1 forwards the packet2 to the NCP4 through the two forwarding paths shown by the thick solid line in FIG. 10 in a round-robin manner according to the drv1 port1.

[0142] After receiving message 2, NCP4 caches message 2 into queue 1 corresponding to drv1 port1; after reordering the messages in queue 1, NCP4 sends the messages in queue 1 to server 4.

[0143] (2) Broadcast entry learning and synchronization within VSI.

[0144] Server 4 and server 5 belong to VSI_1. NCP4 obtains the SysPort ID of port 1 on NCP4 connected to server 4, such as drv1 port1, and obtains the SysPort ID of port 2 on NCP4 connected to server 5, such as drv2 port1. NCP4 adds drv1 port1 and drv2 port1 to the broadcast table corresponding to VSI_1, obtaining the broadcast table entries as shown in Table 5.

[0145] Table 5

[0146] NCP4 can use the fields shown in FIGS. 6a and 6c to extend the BGP EVPN class 3 routing message, synchronize the broadcast table entries shown in Table 5 above to NCP1-NCP3 through the BGP EVPN class 3 routing message, so that NCP1-NCP3 obtain the broadcast table entries shown in Table 5 above, and add the broadcast table entries shown in Table 5 above to the broadcast table corresponding to VSI_1, as shown in Table 6.

[0147] Table 6

[0148] In Table 6, drv3 port1 is the SysPort ID of port 3 on NCP1 connected to server 1. Server 1 belongs to VSI_1.

[0149] After synchronizing the above broadcast table entries, take server 1 sending a message to server 4 as an example. Server 1 sends message 3, and the destination MAC address of message 3 is the MAC address FFF-FFF-FFF of server 4. After receiving message 3, NCP1 does not hit the MAC table entry according to the destination MAC address of message 3. NCP1 obtains each SysPort ID in the broadcast table entries corresponding to VSI_1; for each SysPort ID, according to the SysPort ID, broadcast message 3 through multiple forwarding paths in a round-robin manner.

[0150] (3) Inter-VSI ARP entry learning and synchronization.

[0151] Server 1 belongs to VSI_1, and server 4 belongs to VSI_2.

[0152] The server 4 sends a packet 4 to the NCP 4, the source IP address of the packet 4 is the IP address of the server 4, such as 2.2.2.2, and the source MAC address of the packet 4 is the MAC address of the server 4, such as aaa-bbb-ccc.

[0153] After receiving the packet 4, the NCP 4 obtains the IP address and the MAC address of the server 4, and obtains the SysPort ID of the port 1 connected to the server 4, such as drv1 port1.

[0154] At this time, the NCP 4 learns the ARP table item corresponding to the server 4, as shown in Table 7.

[0155] Table 7

[0156] The NCP 4 can use the fields shown in FIGS. 6a and 6b to extend the BGP EVPN class II route packet, or use the multiplexing MPLS label 2 field shown in FIG. 7 to carry the SysPort ID; and then synchronize the ARP table item shown in Table 7 to the NCP 1-NCP 3 through the BGP EVPN class II route packet, so that the NCP 1-NCP 3 obtains and records the ARP table item shown in Table 7.

[0157] After synchronizing the ARP table item, the NCP 1-NCP 3 can record the ARP table item as shown in Table 7, and record the FIB table item as shown in Table 8.

[0158] Table 8

[0159] Taking that the server 1 sends a packet to the server 4 as an example. The server 1 sends a packet 5, and the destination IP address of the packet 5 is the IP address 2.2.2.2 of the server 4. After receiving the packet 5, the NCP 1 looks up Table 8 and Table 7 according to the destination IP address of the packet 5, and obtains that the out interface is drv1 port1 and the destination MAC address is aaa-bbb-ccc. The NCP 1 forwards the packet 5 to the NCP 4 through the two forwarding paths shown by the thick solid lines in FIG. 10 in a round-robin manner according to aaa-bbb-ccc and drv1 port1.

[0160] After receiving the packet 5, the NCP 4 caches the packet 5 to the queue 1 corresponding to the drv1 port1, and sends the packet in the queue 1 to the server 4 after reordering the packets in the queue 1.

[0161] In the embodiment of the application, the internal processing flow of the NCP 4 for synchronizing the table item to the NCP 1 is shown in FIG. 11.

[0162] The information needing synchronization in the NCP4 includes host routes, static routes, and network segment routes, etc. After the NCP4 learns the ARP table item corresponding to the host routes, the static routes, and the network segment routes, etc., the ARP table item is reported to the BGP module; the BGP module adds the local encapsulation index and the corresponding SysPort ID in the ARP table item, and synchronizes the ARP table item to the NCP1. The SysPort ID is set on the interface by the driver and is managed by the interface management module. The BGP module adds the local encapsulation index and the corresponding SysPort ID in the ARP table item in the manners shown in FIGS. 6a-6d or the manners shown in FIGS. 7-8 to synchronize the ARP table item to the NCP1, and the ARP table item can be carried in the second message.

[0163] When the synchronized information is the host routes and the static routes, the ARP table item carries the next hop IP address and the SysPort ID. When the synchronized information is the network segment routes, the ARP table item corresponding to the silent host is synchronized. In addition, the network segment routes can only carry the SysPort ID.

[0164] For the host routes and the static routes, after the BGP module of the NCP1 receives the ARP table item, the ARP table item is issued to the ARP module, the ARP module generates the host routes according to the ARP table item, and the FIB module is notified to generate the FIB table item.

[0165] For the network segment routes, after the BGP module of the NCP1 receives the ARP table item, the ARP table item is notified to the RM module; the RM module issues the ARP table item to the FIB module, and the FIB module generates the FIB table item.

[0166] Corresponding to the above table item synchronization method, the embodiment of the application further provides a table item synchronization device, as shown in FIG. 12, applied to a first forwarding node, the first forwarding node being in an AI network, the first forwarding node being connected with a host through a first port, and the device comprising:

[0167] A receiving module 121 is configured to receive a first message sent by a host;

[0168] A learning module 122 is configured to learn a first full scheduling table item corresponding to the host according to the first message, the first full scheduling table item including a system port identifier of the first port, and the system port identifier including a device identifier and a port identifier;

[0169] A synchronization module 123 is configured to send a second message to a second forwarding node in the AI network, the second message including the first full scheduling table item, so that the second forwarding node learns a second full scheduling table item, and the second full scheduling table item is the same as the first full scheduling table item.

[0170] In some embodiments, the learning module 122 can be specifically configured to: acquire a source address of the first packet, and acquire a system port identifier of the first port; and learn a first full scheduling table item corresponding to the host according to the acquired address and system port identifier.

[0171] In some embodiments, the second packet includes a tunnel encapsulation TLV, the tunnel encapsulation TLV includes a type field, a length field and a value field; the type field is used to indicate that the value field carries tunnel encapsulation information; the length field is used to carry a length value of the tunnel encapsulation information; and the value field is used to carry the tunnel encapsulation information.

[0172] The value field includes a type subfield, a length subfield and a value subfield; the type subfield is used to indicate that the value subfield carries a system port identifier; the length subfield is used to carry a length value of the system port identifier; and the value subfield is used to carry the system port identifier.

[0173] In some embodiments, the second packet is an EVPN class II route packet, and a network reachable information label 2 field included in the EVPN class II route packet carries the system port identifier.

[0174] In some embodiments, the second packet includes an EVPN class II route packet and an EVPN class V route packet.

[0175] The network reachable information label 2 field included in the EVPN class II route packet carries the system port identifier.

[0176] The gateway address field included in the EVPN class V route packet carries the IP address of the host.

[0177] In some embodiments, the learning module 122 can be specifically configured to: acquire a system port identifier of the first port, the first port being used to receive the first packet; and add the system port identifier to a multicast table corresponding to a virtual switching instance to which the host belongs, to obtain the first full scheduling table item.

[0178] In some embodiments, the second packet includes a tunnel encapsulation TLV, the tunnel encapsulation TLV includes a type field, a length field and a value field; the type field is used to indicate that the value field carries tunnel encapsulation information; the length field is used to carry a length value of the tunnel encapsulation information; and the value field is used to carry the tunnel encapsulation information.

[0179] The value field includes a type subfield, a length subfield and a value subfield; the type subfield is used to indicate that the value subfield carries a plurality of system port identifiers; the length subfield is used to carry a length value of the plurality of system port identifiers; and the value subfield is used to carry the plurality of system port identifiers.

[0180] The value subfield includes a length sub-subfield and at least one value sub-subfield; the length sub-subfield is used to indicate the number of system port identifications; and the value sub-subfield is used to carry one system port identification.

[0181] In some embodiments, the type subfield is also used to indicate that the value subfield carries an encapsulation index; the length subfield is also used to carry a length value of the encapsulation index; and the value subfield is also used to carry the encapsulation index.

[0182] In the technical scheme provided by the embodiments of the present application, the forwarding node carries the full scheduling table item in a message and sends it to other forwarding nodes, so that the other forwarding nodes learn the same full scheduling table item, i.e., the system port identification of the connected port when the host accesses the forwarding node is included in the full scheduling table item. In the embodiments of the present application, the synchronization of the full scheduling table item is realized by transmitting messages between the forwarding nodes, and the controller is not needed to synchronize the full scheduling table item, thereby reducing the influence of the synchronization of the table item on the control plane and improving the performance and expansion capability of the control plane.

[0183] The embodiments of the present application further provide a network device, as shown in FIG. 13, which includes a processor 131 and a machine readable storage medium 132, the machine readable storage medium 132 stores machine executable instructions which can be executed by the processor 131, and the processor 131 is prompted by the machine executable instructions to implement any of the above-mentioned table item synchronization methods.

[0184] The machine readable storage medium can include a random access memory (RAM) and can also include a non-volatile memory (NVM), for example, at least one disk memory. Optionally, the machine readable storage medium can also be at least one storage device located away from the aforementioned processor.

[0185] The processor can be a general purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component.

[0186] In a further example provided in the present application, a machine readable storage medium is provided, and the machine readable storage medium stores a computer program, and the computer program is executed by a processor to implement the entry synchronization method described above.

[0187] In a further example provided in the present application, a computer program is provided, and the computer program is executed by a processor to implement the entry synchronization method described above.

[0188] In the above examples, all or part of the embodiments can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present application are generated. The computer can be a general purpose computer, a special purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another computer readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) mode. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with one or more available media. The available media can be magnetic media (for example, floppy disk, hard disk, magnetic tape), optical media (for example, DVD), or semiconductor media (for example, solid state disk (SSD)) and the like.

[0189] It should be noted that in this paper, the relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitation, the element defined by the statement "including a" does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0190] Each of the embodiments in the specification is described in a relevant manner, and the same or similar parts between the embodiments can be referred to each other. Each of the embodiments focuses on the difference from other embodiments. In particular, for the device, network equipment, storage medium and computer program embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiment.

[0191] The above only describes the preferred embodiments of the present application, and is not used to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for synchronizing entries, the method comprising: The method is applied to a first forwarding node in an artificial intelligence (AI) network, the first forwarding node is connected to a host through a first port, and the method comprises the following steps: receiving a first packet sent by the host; learning a first full scheduling table item corresponding to the host according to the first packet, wherein the first full scheduling table item comprises a system port identifier of the first port, and the system port identifier comprises a device identifier and a port identifier; sending a second packet to a second forwarding node in the AI network, wherein the second packet comprises the first full scheduling table item, so that the second forwarding node learns a second full scheduling table item, and the second full scheduling table item is the same as the first full scheduling table item.

2. The method of claim 1, wherein, The method of learning the first full scheduling table item corresponding to the host according to the first packet specifically comprises the following steps: obtaining a source address included in the first packet, and obtaining a system port identifier of the first port; learning the first full scheduling table item corresponding to the host according to the obtained source address and system port identifier.

3. The method of claim 2, wherein, The second packet comprises a tunnel encapsulation TLV, and the tunnel encapsulation TLV comprises a type field, a length field and a value field; the type field is used to indicate that the value field carries tunnel encapsulation information; the length field is used to carry a length value of the tunnel encapsulation information; and the value field is used to carry the tunnel encapsulation information; the value field comprises a type subfield, a length subfield and a value subfield; the type subfield is used to indicate that the value subfield carries one system port identifier; the length subfield is used to carry a length value of the one system port identifier; and the value subfield is used to carry the one system port identifier.

4. The method of claim 2, wherein, The second packet is an EVPN class II route packet, and a network reachable information label 2 field included in the EVPN class II route packet carries the system port identifier.

5. The method of claim 2, wherein, The second packet comprises an EVPN class II route packet and an EVPN class V route packet; the network reachable information label 2 field included in the EVPN class II route packet carries the system port identifier; the gateway address field included in the EVPN class V route packet carries an IP address of the host.

6. The method of claim 1, wherein, The method of learning the first full scheduling table item corresponding to the host according to the first packet specifically comprises the following steps: obtaining a system port identifier of the first port used for receiving the first packet; adding the system port identifier to a multicast table corresponding to a virtual switching instance to which the host belongs, to obtain the first full scheduling table item corresponding to the host. The second packet comprises a tunnel encapsulation TLV, and the tunnel encapsulation TLV comprises a type field, a length field and a value field; 7. The method of claim 6, wherein, the type field is used to indicate that the value field carries tunnel encapsulation information; the length field is used to carry a length value of the tunnel encapsulation information; and the value field is used to carry the tunnel encapsulation information; the value field comprises a type subfield, a length subfield and a value subfield; the type subfield is used to indicate that the value subfield carries a plurality of system port identifiers; the length subfield is used to carry length values of the plurality of system port identifiers; and the value subfield is used to carry the plurality of system port identifiers; ​ The value subfield includes a length sub-subfield and at least one value sub-subfield; the length sub-subfield is used to indicate the number of system port identifications; and the value sub-subfield is used to carry one system port identification.

8. The method according to claim 3 or 7, characterized in that, The type subfield is also used to indicate that the value subfield carries an encapsulation index; the length subfield is also used to carry a length value of the encapsulation index; and the value subfield is also used to carry the encapsulation index.

9. A table entry synchronization apparatus, characterized by comprising: The device is applied to a first forwarding node in an artificial intelligence (AI) network, and the first forwarding node is connected to a host through a first port. The receiving module is configured to receive a first packet sent by the host. The learning module is configured to learn a first full scheduling table item corresponding to the host according to the first packet, the first full scheduling table item including a system port identification of the first port, and the system port identification including a device identification and a port identification. The synchronization module is configured to send a second packet including the first full scheduling table item to a second forwarding node in the AI network, so that the second forwarding node learns a second full scheduling table item identical to the first full scheduling table item.

10. The apparatus of claim 9, wherein, The learning module is specifically configured to obtain a source address included in the first packet and obtain a system port identification of the first port, and learn the first full scheduling table item corresponding to the host according to the obtained source address and system port identification.

11. The apparatus of claim 10, wherein, The second packet includes a tunnel encapsulation TLV, and the tunnel encapsulation TLV includes a type field, a length field and a value field. The type field is used to indicate that the value field carries tunnel encapsulation information. The length field is used to carry a length value of the tunnel encapsulation information; and the value field is used to carry the tunnel encapsulation information. The value field includes a type subfield, a length subfield and a value subfield; the type subfield is used to indicate that the value subfield carries one system port identification; the length subfield is used to carry a length value of the one system port identification; and the value subfield is used to carry the one system port identification. The second packet is an EVPN class II route packet, and a network reachable information label 2 field included in the EVPN class II route packet carries the system port identification.

12. The apparatus of claim 10, wherein, The second packet includes an EVPN class II route packet and an EVPN class V route packet.

13. The apparatus of claim 10, wherein, The network reachable information label 2 field included in the EVPN class II route packet carries the system port identification. A gateway address field included in the EVPN class V route packet carries an IP address of the host. The learning module is specifically configured to obtain a system port identification of the first port used to receive the first packet, and add the system port identification to a multicast table corresponding to a virtual switching instance to which the host belongs to obtain the first full scheduling table item.

14. The apparatus of claim 9, wherein, The second packet includes a tunnel encapsulation TLV, and the tunnel encapsulation TLV includes a type field, a length field and a value field.

15. The apparatus of claim 14, wherein, The type field is used to indicate that the value field carries tunnel encapsulation information. ​ The length field is used to carry a length value of the tunnel encapsulation information; and the value field is used to carry the tunnel encapsulation information. The value field comprises a type subfield, a length subfield and a value subfield; the type subfield is used to indicate that the value subfield carries a plurality of system port identifications; the length subfield is used to carry a length value of the plurality of system port identifications; and the value subfield is used to carry the plurality of system port identifications. The value subfield comprises a length sub-subfield and at least one value sub-subfield; the length sub-subfield is used to indicate a number of the plurality of system port identifications; and the value sub-subfield is used to carry one system port identification.

16. The apparatus of claim 11 or 15, wherein, The type subfield is also used to indicate that the value subfield carries an encapsulation index; the length subfield is also used to carry a length value of the encapsulation index; and the value subfield is also used to carry the encapsulation index.

17. A network device, comprising: The machine readable storage medium stores a computer program which can be executed by the processor, and the processor is caused by the computer program to implement the method steps of any one of claims 1-8.

18. A machine-readable storage medium, characterized in that, The machine readable storage medium stores a computer program which can be executed by the processor, and the processor is caused by the computer program to implement the method steps of any one of claims 1-8.

19. A computer program, characterized in that, The computer program is executed by the processor to implement the method steps of any one of claims 1-8.

Images