A fast switching method for ring network link failure based on single ring

Abstract

The invention discloses newly added port attributes, which include Left, Right and Local. The ring network member ports are set to Right and Left attributes in a cross order, and the non-ring network member ports are all set to Local attributes. The FDB table item dest portMask is adapted according to the port attribute. The Ethernet switch detects whether the loop exists by means of a specific protocol. If the link exists, the MAC learned by the ring network member port will automatically add another member port. When a link in the ring network is linkdown, the switches on both sides of the broken link immediately switch the Left or Right forwarding port to notify other switches in the ring network of the link break. After receiving the linkdown event, the remaining switches clear the Left+Right attributes in the FDB table item and set them to none, and the FDB table item enters a waiting convergence state. The FDB table item is converged. After the convergence is completed, the mutual communication between EPs is forwarded unidirectionally according to the FDB table item. The invention can make it unnecessary to refresh the FDB table during the ring network switching process, thereby avoiding the FDB flush performance consumption, bandwidth waste and security problems caused by message flood.

Description

Technical Field

[0001] The invention relates to the technical field of Ethernet ring networks, and in particular to a ring network link failure fast switching method based on a single ring. Background Art

[0002] In order to perform link backup and improve network reliability in Ethernet switching networks, ring network redundant links are usually used. Currently, common Ethernet ring network redundancy is mainly divided into the following types: ring network based on spanning tree STP / RSTP / MSTP protocol, ring network based on private ring network protocol, and ring network based on ERPS (G.8032) universal ring network protocol.

[0003] For the existing single-ring network link, such as Figure 1 As shown,

[0004] Figure 1 The basic strategy of the central ring network redundancy protocol is as follows:

[0005] (1) In the initial state, all links are normal and a physical loop exists. Actively block one of the links (set the ports on both sides or one side of the link to the block state). At this point, there is no loop in the logical network topology. The MAC in the FDB table is bound to a unique port, and the message is forwarded according to the FDB table.

[0006] (2) When a link on the loop goes down, the event is notified by means of a message. The blocking port is switched to the forwarding state. At the same time, the FDB flush operation is performed on all ports on the loop, and learning and refreshing are performed in subsequent communications.

[0007] (3) When the link is restored, the switch on the ring confirms the linkup event with the help of protocol messages. It will reselect a port or directly use the original blocked port for blocking. At the same time, it will perform an FDBflush operation on all member ports on the ring.

[0008] After the network topology is changed, ERPS and other ring network redundancy protocols need to refresh the FDB entries on the ring network ports, flush the FDB entries under the ring network member ports, and relearn convergence. This process will test the FDB flushing capability of the switch; at the same time, unknown messages will be flooded, wasting bandwidth and introducing security issues (sensitive data). Summary of the invention

[0009] In view of the deficiencies in the prior art, the present invention provides a single-ring based ring network link failure fast switching method, which eliminates the need to refresh the FDB table during the ring network switching process, thereby avoiding FDB flush performance consumption, bandwidth waste and security issues caused by message flooding.

[0010] The present invention is achieved through the following technical solutions:

[0011] A method for fast switching of a ring network link failure based on a single ring, comprising the following steps:

[0012] Added port attributes, including Left, Right, and Local. Ring member ports are set to Right and Left attributes in the crossover order, and non-ring member ports are set to Local attributes.

[0013] The FDB table entry dest portMask is adapted according to the port attributes;

[0014] The Ethernet switch uses a specific protocol to detect whether a loop exists. If a link exists, the MAC learned by the ring member port will automatically add another member port;

[0015] When a link in the ring network is linkdown, the switches on both sides of the link break immediately switch the Left or Right forwarding port to notify other switches in the ring network of the link break. After receiving the linkdown event, the remaining switches clear the Left+Right attributes in the FDB table entry and set them to none. The FDB table entry enters the waiting convergence state.

[0016] Converge the FDB entries;

[0017] After convergence is completed, communication between EPs is forwarded unidirectionally according to the FDB entries.

[0018] Furthermore, the step of: converging the FDB table entries specifically includes the following steps:

[0019] When the Left or Right port receives a message, it checks whether the FDB entry exists:

[0020] If it does not exist, learn it and set the Left or Right port corresponding to the receiving port;

[0021] If it exists but none is set, it is updated to the Left or Right port.

[0022] If it exists, but the Left or Right valid bit is set, the FDB entry remains unchanged.

[0023] Furthermore, the step: after convergence is completed, the mutual communication between EPs is forwarded according to the FDB table entry; and then the following steps are also included:

[0024] When the linkdown link is restored, the switches on both sides of the link restoration immediately update the FDB table entry with dest portMask as Left, Right or None, and at the same time, notify the other switches in the ring network of the link restoration event;

[0025] When other switches in the ring network receive the re-link event, they immediately update the FDB entry with dest portMask as Left, Right or None and set dest portMask to Left+Right.

[0026] Furthermore, the step: the FDB table item dest portMask is adapted according to the port attribute, specifically including:

[0027] The link is established. At this time, the FDB is empty. The message will be flooded in the network and learned. The Local port will set the FDB table item dest portMask to the corresponding port. The Left or Right port will set the FDB table item dest portMast to the Left or Right port.

[0028] Furthermore, for the FDB entry that has entered the pending convergence state, corresponding messages are allowed to be forwarded on both the Left and Right ports at the same time.

[0029] Furthermore, for the FDB table entry that enters the waiting-for-convergence state, the ring network member port does not forward the message sent by the switch or the downstream EP after receiving it.

[0030] Compared with the prior art, the advantages of the present invention are:

[0031] 1. This device adds FDB table attributes and specific forwarding strategies, so that there is no need to flush FDB during ring network oscillation. Only the dest portMask attribute table needs to be updated (completed by hardware), which greatly shortens the ring network convergence time (no more than 5ms). At the same time, since the loop FDB table entries always exist, the bandwidth waste caused by a large number of message floods before convergence and the leakage of sensitive data are avoided. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a typical single-ring network scenario diagram;

[0033] Figure 2 A flowchart of a method for fast switching of a ring network link failure based on a single ring;

[0034] Figure 3 It is a schematic diagram of the structure of a typical single-ring network topology;

[0035] Figure 4 This is the state diagram of the single-ring network topology link initialization;

[0036] Figure 5 This is a state diagram of the single-ring network topology FDB table entry waiting for convergence;

[0037] Figure 6 This is the state diagram when the FDB table entry convergence of the single-ring network topology is completed;

[0038] Figure 7 This is the state diagram when the single-ring network topology link returns to normal. DETAILED DESCRIPTION

[0039] The following is a further non-restrictive detailed description of the technical solution of the invention in conjunction with the preferred embodiments and the accompanying drawings. In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate the orientation or position relationship based on the orientation or position relationship shown in the accompanying drawings. In addition, the terms "first" and "second" are used only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and cannot be understood as limitations on the present invention.

[0040] like Figure 2 As shown, a method for fast switching of a ring network link failure based on a single ring according to an embodiment of the present invention comprises the following steps:

[0041] S1: Single-ring link initialization phase:

[0042] S10: Add new port attributes, including Left, Right and Local. Ring network member ports are set to Right and Left attributes in the cross order, and non-ring network member ports are set to Local attributes.

[0043] refer to Figure 3The ring network link of a single ring has at least one Ethernet switch. The port attributes of two adjacent Ethernet switches are Left and Right respectively. The port attribute of the EP connecting to the Ethernet switch is Local. In addition, the message is transmitted according to Figure 3 The packets are forwarded in the direction described in the figure, that is, they are all forwarded from the Left port.

[0044] S11: The FDB table entry dest portMask is adapted according to the port attributes.

[0045] In this document, dest portMask refers to the destination port bitmap, which is used to represent multiple ports. For example, 0x1 represents port0, and 0x5 represents port0 and port2.

[0046] Specifically, the FDB entry learned by the Local port only contains this port in the dest portMask; the FDB entry learned by the Left or Right port initially only contains itself in the dest portMask, but once an event is detected in the ring network, the Right or Left port will be automatically added to the dest portMask. The FDB entry destportMask corresponding to the Left and Right ports is allowed to be set to the none state. That is: when the local attribute port performs SA learning, only the bit corresponding to the port in portMask is set. When the ring network member port (left or right attribute) performs SA learning, the port that receives the message is first added to the portMask, and the subsequent DA matching message is forwarded according to the portMask; when a ring network event is detected, another ring network member port is automatically added to the portMask, and the subsequent DA matching message is forwarded from the Left port according to the default rule. The FDB entry corresponding to the ring network member port allows the dest portMask to be 0, and the subsequent DA matching message is forwarded to both ports at the same time.

[0047] Specifically: the link is established, the FDB is empty, the message will be flooded in the network and learned; the Local port will set the FDB table item dest portMask to the corresponding port; the Left or Right port will set the FDB table item destportMast to the Left or Right port.

[0048] S12: The Ethernet switch uses a specific protocol to detect whether a loop exists. If a link exists, the MAC learned by the ring network member port will automatically add another member port;

[0049] Assuming that all EPs interact, we will eventually establish Figure 4 The FDB table entry shown

[0050] In the figure, MAC1 is the MAC belonging to EP1, and so on; sw1-sw4 are 4 Ethernet switches.

[0051] In this state, a typical communication example is as follows:

[0052] (1) Example: EP1 sends a message to EP2. sw1 queries FDB and finds that the MAC2 entry is Local, so it forwards it directly to destport.

[0053] (2) EP1 sends a message to EP6. sw1 queries FDB and finds that the MAC6 entry is Left+Right, so it forwards it to the Left port. After sw2 receives the message, it finds that the MAC6 entry is Left+Right and continues to forward it to the Left port. sw3 receives the message and queries FDB. It finds that the MAC6 entry is Local and forwards it directly to the dest port.

[0054] Among them, EP1 sends out messages that need to be flooded (such as bcast, uncast, mcast / ucast, etc.), and switch1-4 forwards them from the Left port except for flooding in the Local port. After sw1 receives the message from the Right port, it automatically discards the message that is confirmed to be sent by the local device through the detection of SA filter.

[0055] S2: Link failure stage:

[0056] S20: When a link in the ring network is linkdown, the switches on both sides of the link break immediately switch the Left or Right forwarding port to notify other switches in the ring network of the link break. After receiving the linkdown event, the remaining switches clear the Left+Right attributes in the FDB table entry and set them to none. The FDB table entry enters the pending convergence state.

[0057] For the FDB entry that enters the pending convergence state, corresponding messages are allowed to be forwarded on both the Left and Right ports at the same time.

[0058] like Figure 5 As shown in the figure, if the link between sw2 and sw3 is linkdown, the FDB table entry enters the waiting convergence state after the event notification is completed. In this state, a typical communication example is as follows:

[0059] (1) EP3 sends a message to EP4. sw2 queries FDB and finds that the MAC4 entry is Local, so it forwards it directly to destport.

[0060] (2) EP3 sends a message to EP6. sw2 queries FDB and finds that the MAC6 entry is Right, so it forwards it to the Right port. After sw1 and sw4 receive the message in turn, they find that the MAC6 entry is none. Since it is received from the Left port, it only forwards it to the Right port. Finally, sw3 receives the message and forwards it to the Local port. EP6 sends a message to EP3 in a similar way, but the path is reversed.

[0061] (3) EP1 sends a message to EP6. sw1 queries FDB and finds that the MAC6 entry is none. It then forwards the message to both the Left and Right ports. After sw2 receives the message, it finds that the MAC6 entry is Right and discards the message. sw4 receives the message and queries FDB. It finds that the MAC6 entry is none. Since the message is received from the Left port, it only forwards the message to the Right port. Finally, sw3 receives the message and forwards it to the Local port.

[0062] Among them, EP1 sends out messages that need to be flooded (such as bcast, uncast, mcast / ucast, etc.), and sw1-4 will flood the remaining ports except the receiving port.

[0063] S21: Converging the FDB entries.

[0064] Specifically: When the Left or Right port receives a message, it determines whether the FDB entry exists:

[0065] If it does not exist, learn it and set the Left or Right port corresponding to the receiving port;

[0066] If it exists but none is set, it is updated to the Left or Right port.

[0067] If it exists, but the Left or Right valid bit is set, the FDB entry remains unchanged.

[0068] S22: Convergence is completed, and the communication between EPs is forwarded unidirectionally according to the FDB table entries.

[0069] Assuming the link between sw2 and sw3 is linkdown, the convergence completion status is as follows: Figure 6 As shown, in this state, a typical communication example is as follows:

[0070] The communication between EPs is forwarded according to the FDB table information.

[0071] Among them, for the messages that need to be flooded (such as bcast, uncast, mcast / ucast, etc.), sw1-4 will flood the remaining ports except the receiving port.

[0072] S3: Link recovery phase:

[0073] S30: When the linkdown link is restored, the switches on both sides of the link restoration immediately update the FDB table entry with dest portMask as Left, Right or None, and at the same time, notify the link restoration event to other switches in the ring network.

[0074] S31: When other switches in the ring network receive the re-link event, they immediately update the FDB table entry whose dest portMask is Left, Right or None, and set the dest portMask to Left+Right.

[0075] When restoring, the dest portMask is set to Left+Right.

[0076] The final result is Figure 7 As shown in the figure, the link is restored to the normal state. Subsequent communications will be carried out in accordance with the normal link state.

[0077] Regarding the single-ring-based ring network link failure fast switching method of the present invention, it should be noted that:

[0078] (1) When there is only one valid member port in the FDB entry dest portMask, the message is forwarded according to destportMask. This mechanism is applicable to local ports, ring network initialization phase, and after the ring network is disconnected and converged.

[0079] (2) When there are two valid member ports in the FDB entry dest portMask, the message is forwarded in the direction shown in the above figure, that is, all are forwarded from the Left port. This mechanism is for cross-switch communication scenarios when a ring network exists.

[0080] (3) When there is no valid member port in the FDB entry dest portMask, the message will be forwarded on both the Left and Right ports. At the same time, the source port filter mechanism is still effective. This mechanism is designed for scenarios where the ring network is disconnected and has not yet converged.

[0081] (4) Port blocking is not performed in the ring network link, and SA filter is used to prevent flood message loopback. That is, after receiving the message sent by the switch or the downstream EP on the ring network member port, it is not forwarded.

[0082] The present invention increases FDB table item attributes and a specific forwarding strategy, so that during the ring network oscillation, there is no need to perform FDB flush, and only the dest portMask attribute table update (completed by hardware) needs to be completed, which greatly shortens the ring network convergence time (no more than 5ms). At the same time, since the loop FDB table items always exist, bandwidth waste caused by a large number of message floods before convergence and leakage of sensitive data are avoided.

[0083] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A method for fast switching of a ring network link failure based on a single ring, characterized in that: The following steps are involved: Added port attributes, including Left, Right and Local. Ring member ports are set to Right and Left attributes in the crossover order, and non-ring member ports are set to Local attributes. Among them, the message is forwarded from the Left port. The FDB table entry dest portMask is adapted according to the port attributes; The Ethernet switch uses a specific protocol to detect whether a loop exists. If a link exists, the MAC learned by the ring member port will automatically add another member port; When a link in the ring network is linkdown, the Left+Right attributes of the switches on both sides of the link are immediately switched to the normally usable Left or Right forwarding port, notifying other switches in the ring network of the link down. After receiving the linkdown event, the remaining switches clear the Left+Right attributes in the FDB table entry and set them to none, and the FDB table entry enters the waiting-for-convergence state. For the FDB table entry that enters the waiting-for-convergence state, the corresponding message is allowed to be forwarded on both the Left and Right ports at the same time. Converge the FDB entries; After convergence is completed, communication between EPs is forwarded unidirectionally according to the FDB entries.

2. The method for fast switching of a ring network link failure based on a single ring according to claim 1, characterized in that: The step of converging the FDB table entries specifically includes the following steps: When the Left or Right port receives a message, it checks whether the FDB entry exists: If it does not exist, learn it and set the Left or Right port corresponding to the receiving port; If it exists, but none is set, it is updated to the Left or Right port; If it exists, but the Left or Right valid bit is set, the FDB entry remains unchanged.

3. The method for fast switching of a ring network link failure based on a single ring according to claim 1, characterized in that: The step: after convergence is completed, the mutual communication between EPs is forwarded according to the FDB table entry; and then the following steps are also included: When the linkdown link is restored, the switches on both sides of the link restoration immediately update the FDB table entry with dest portMask as Left, Right or None, and at the same time, notify the other switches in the ring network of the link restoration event; When other switches in the ring network receive the re-link event, they immediately update the FDB entry with dest portMask as Left, Right or None and set dest portMask to Left+Right.

4. The method for fast switching of a ring network link failure based on a single ring according to claim 1, characterized in that: The step: the FDB table item dest portMask is adapted according to the port attribute, specifically including: The link is established. At this time, the FDB is empty. The message will be flooded in the network and learned. The Local port will set the FDB table item dest portMask to the corresponding port. The Left or Right port will set the FDB table item dest portMast to the Left or Right port.

5. The method for fast switching of a ring network link failure based on a single ring according to claim 1, characterized in that: For FDB entries that are in the pending convergence state, the ring network member ports do not forward the packets received from the switch or the downstream EP.

Images