CN1848767A - Filtering frames for link aggregation marker protocol - Google Patents

Abstract

Cleanup mechanism in link aggregation group management. In order to be able to quickly move sessions between links in a link aggregation group, network elements employ clearing mechanisms. A network element implementing this purge mechanism may inhibit the allocation of additional frames to the output queue associated with the aggregated port, and potentially drop some or all frames from the output queue associated with the aggregated port. In conjunction with the dropping of the frame, the network element may exchange one or more flag messages and flag responses with the remote network element. After receiving an appropriate response, the network element can resume distributing frames to the affected ports.

Description

Clearing Mechanism in Link Aggregation Group Management

technical field

The present invention generally relates to network management, and more particularly relates to a purge mechanism in link aggregation group management.

Background technique

A Link Aggregation Group (LAG) combines multiple physical network links into a single logical link that provides aggregated throughput and high availability to terminals. Communication between the two terminals takes place over this single logical link LAG.

Contents of the invention

According to the present invention, techniques are provided for moving sessions between links using a clearing mechanism.

According to a specific embodiment, a method for redistributing sessions among links in a link aggregation group aggregates a plurality of physical links connected between a first network element and a second network element into a link Groups are aggregated and received frames, associated with one or more sessions, are distributed among output queues corresponding to the respective physical links. The method decides to move at least one of said sessions between said physical links; inhibits allocation of additional received frames to output queues corresponding to said physical links; clearing a frame; and transmitting a flag message on said physical link. The method receives a flag response and, responsive to receipt of the flag response, enables distribution of subsequently received frames among physical links.

Embodiments of the present invention provide various technical advantages. Particular embodiments provide an efficient mechanism for moving communications between links using the Link Aggregation Labeling Protocol. For example, contrary to the traditional solution that takes a long time to move communication, the combination flag protocol provides a clearing mechanism, which can more efficiently and effectively move communication between terminals. According to a particular embodiment, the clearing mechanism also ensures that frames are not rearranged or duplicated when moving communication from one link to another. According to a particular embodiment, link availability may be increased using a purge mechanism in conjunction with a flagging protocol. This allows for continuous communication without interruption while moving communication between links.

Those skilled in the art will easily understand other technical advantages of the present invention from the following drawings, descriptions and claims. Additionally, while specific advantages may be enumerated, different embodiments may include all, some, or none of the enumerated advantages.

Description of drawings

In order to understand the present invention and its advantages more completely, now in conjunction with accompanying drawing, carry out following explanation, in accompanying drawing:

FIG. 1 shows a communication system including a LAG implementing a clearing mechanism according to a specific embodiment of the present invention;

Figure 2 illustrates exemplary network elements in the system of Figure 1;

Fig. 3 is a flow chart, and it represents the method that uses the frame allocator at the transmission network element place to move session between ports by clearing mechanism;

Fig. 4 is a flow chart, and it represents the method that uses the frame collector at the receiving network element place to respond to the decision of moving session between ports through clearing mechanism;

Figure 5 is a flowchart representing a method of implementing an allocation adjustment mechanism for a LAG;

Fig. 6 is a flow chart, and it represents the method that uses the frame allocator at the transmission network element place to realize special flag message; And

Figure 7 is a flow diagram representing a method of responding to a specific flag message using a frame collector at a receiving network element.

Detailed ways

Figure 1 shows a communication system, indicated generally at 10, including a LAG implementing a purge mechanism. Terminals 18 communicate with each other over network 12 using network elements 16 . Typically, network elements 16 are arranged to form a LAG for high-speed communication between terminals 18 . To support the operation of the LAG, network element 16 may implement the following techniques, including: clearing mechanisms for quickly moving sessions between links in the LAG, moving traffic efficiently from links in the LAG that fail due to faults or other reasons The Extended Labeling Protocol, and the Allocation Adjustment Algorithm that help to efficiently support and fully utilize the links in the LAG. Network element 16 may implement some or all of these techniques to support the operation of the LAG.

Network 12 represents a communication device including hardware and any suitable control logic for interconnecting elements connected to network 12 and facilitating communication between terminals 18 . Network 12 may include a local area network (LAN), a metropolitan area network (MAN), any other public or private network, logical, regional or global communication network, an intranet, other suitable wired or wireless communication links, or any combination of the foregoing. Additionally, network 12 may include gateways, routers, hubs, switches, and any other combination of hardware, software, or a combination of the foregoing that may implement any suitable protocol or communication.

In the illustrated embodiment, network 12 includes at least one network manager 14 and a plurality of network elements 16 . Network manager 14 monitors and controls the behavior of network elements 16 . For example, network manager 14 provides configuration information for network elements 16 . Specifically, network manager 14 may form and manage LAGs between network elements 16 . As an example, network manager 14 may monitor traffic within network 12 and alter the use of links 17 and LAGs in response to network conditions.

Network element 16 represents network communication equipment including appropriate control logic that facilitates communication between terminals 18 . For example, network elements 16 may include switches, routers, gateways, servers, or other suitable network devices. According to a particular embodiment, network elements 16 communicate with each other via high speed electrical signals. In the illustrated embodiment, LAG 15 is formed between network elements 16a and 16b to provide increased bandwidth and increased availability during communications. According to a particular embodiment, a network element 16a negotiates a LAG 15 with another link aggregation capable network element 16b. To form a LAG 15, one or more physical links 17 between network elements 16 are aggregated together.

Each link 17 represents any suitable channel capable of exchanging signals between network elements 16 . Network element 16 may conduct multiple communications simultaneously over multiple links 17 . Communications may be moved between links 17 as communications are performed. Terminal 18 treats LAG 15 comprising one or more physical links 17 as a single logical link for communication. Network elements 16 may be aggregated in any suitable manner, and any suitable number of links 17 may be aggregated together to form one or more LAGs 15. For example, a network element 16 may have a total of eight links 17, aggregate three links 17 to form a first LAG 15, aggregate another two links 17 to form a second LAG 15, and operate the remaining three links 17 individually Does not aggregate.

Terminal 18 represents any suitable device that can communicate with network 12 . Communication between terminals 18 occurs by exchanging frames. Terminals 18 exchange audio, voice, data, video or other information within system 10 using any suitable communication protocol. Terminal 18 may be any combination of hardware and/or software that provides communication services to users. For example, terminals 18 include servers, personal computers (eg, laptop or desktop computers), Internet Protocol (IP) telephones, or any suitable device that can communicate within system 10 .

According to a particular embodiment, components within system 10 use the Ethernet standard for frame communication. A frame includes any suitable segment of data, such as a packet, frame or cell. Additionally, Ethernet and Ethernet standards include communication protocols developed to handle the transmission of frames between components, including any extensions, additions, and/or further developments that arise to these protocols. For example, Ethernet standards include protocols set forth in Institute of Electrical and Electronics Engineers (IEEE) 802.3 and appendices.

As mentioned above, LAG 15 acts as a single logical link formed from a plurality of individual physical links 17 connected between network elements 16. During operation, two network elements 16 connected by a particular LAG 15 may treat that LAG 15 as a single physical connection, potentially with some limitations. As an operational example, assume that terminal 18a communicates with terminal 18b over network 12, and that links 17a-17c between network elements 16a and 16b aggregate to form LAG 15. Communication between network element 16a and network element 16b may be referred to as a session. According to a particular embodiment, network element 16 maintains each session on a single link 17 within a given LAG 15. This can help keep frames ordered within a session. If sessions are distributed unevenly across links 17 in LAG 15, this results in poor utilization of the full bandwidth of LAG 15. Additionally, the failure of one link 17 would potentially cut off sessions over that link 17 . Thus, sessions among links 17 within LAG 15 may be switched in response to link failure, poor link utilization, reconfiguration, or other suitable circumstances.

During operation, network element 16 may distribute received frames among links 17 of LAG 15 using any suitable technique. According to a particular embodiment, the network element 16 employs an allocation algorithm to select a particular link 17 for each received frame. For example, network element 16 may select a particular one of links 17 within LAG 15 based on addressing information (eg, source or destination address information) contained in each frame. This algorithm can ensure that all frames from one terminal 18 to another terminal 18 are transmitted along the same link 17 and thus can ensure correct ordering of the frames. This allocation algorithm does not require a state-based memory to track the allocation of sessions, but results in poor session allocation between links 17 . As an alternative, sessions may be assigned to links 17 in LAG 15 in a round-robin manner. However, using state-based allocation techniques such as round-robin techniques requires memory operations since the allocation of sessions between different links 17 has to be tracked.

In order to reduce the need for state-based allocation techniques while gaining the advantage of link 17 utilization uniformity, network element 16 may support a mechanism for adjusting frame allocation among links 17 in LAG 15. For example, network element 16 may change the allocation of communications between links 17 in a particular LAG 15, assuming that one or more links 17 in that LAG 15 are underutilized. According to particular embodiments, network element 16 supports a plurality of different allocation algorithms, and network manager 14 may select between these different algorithms in response to any appropriate network status. For example, network elements 16 may each provide a plurality of different algorithms, each of which calculates a particular link 17 in LAG 15 based on some combination of source and/or destination address information. By using different combinations and parts of the addressing information and potentially applying different functions, these algorithms can make different allocations of frames between links 17 while still maintaining correct frame ordering. Network manager 14 may automatically or manually change the allocation algorithm used by one or more network elements 16 to overcome underutilization of LAG 15.

In addition to providing multiple allocation algorithms, network element 16 may also support the use of allocation parameters in conjunction with one or more allocation algorithms. These parameters can also affect the allocation function obtained by applying the allocation algorithm. For example, allocation parameters may change the portion of addresses considered by a particular allocation algorithm. Used in combination, a relatively small number of allocation algorithms and parameters can provide a large number of potential allocation functions.

Sessions may be moved between links 17 in the LAG 15 in the event of changes to the allocation algorithm or parameters, link 17 failure, LAG 15 reconfiguration, or other appropriate circumstances. To quickly move traffic between links 17 in LAG 15, network element 16 may employ a clearing mechanism. Alternatively or additionally, network element 16 may employ an extended marker protocol in the event of link 17 failure or failure.

In order to normally move communications between links 17 within LAG 15, network element 16 may support a flagging protocol, which may be based on a standard such as Institute of Electrical and Electronics Engineers (IEEE) 802.3 Clause 43. Continuing with the above example, assume that a session between terminal 18a and terminal 18b involves passing a stream of frames from terminal 18a to terminal 18b, and that network element 16a transmits these frames to network element 16b using link 17a. In response to link 17a failure, session reallocation, or other appropriate circumstances, network element 16a may determine to move the session on link 17a to another link 17 in LAG 15.

In order to be able to move the session (or sessions) quickly, network element 16a may use a clearing mechanism. In an exemplary embodiment, the purge mechanism includes inhibiting the allocation of additional frames to the output queue associated with link 17a and potentially dropping some or all frames from the output queue associated with link 17a. Network element 16a sends a message about the mobile session to network element 16b. For example, network element 16a may send a flag message to network element 16b using an administrative queue associated with link 17a. When network element 16b responds to the message, network element 16a may move the session to another link 17 within LAG 15. The use of flag messages and flag responses can help ensure that frames transmitted on LAG 15 are properly sequenced. Sessions can be quickly moved between links 17 by employing a purge mechanism.

According to a particular embodiment, network element 16a moves all sessions occurring on link 17a using this purge mechanism. For example, network element 16a may move multiple sessions from link 17a to link 17c, or may propagate sessions between two or more other links 17 within a given LAG 15. Additionally, network element 16 may use the clearing mechanism on multiple different links 17 simultaneously. For example, network element 16 may use a purge mechanism on all links 17 in LAG 15 in conjunction with session reassignment between links 17 based on changes in the allocation algorithm.

Among other things or under other circumstances, network element 16 may implement an extended flag protocol to further facilitate supporting communications between mobile links 17 . For example, if one of the links 17 within the LAG 15 fails or fails, the network element 16 may employ the extended flag protocol. According to a particular embodiment, a network element 16 may respond to a failure or failure of a particular link 17 within the LAG 15 by exchanging a flag-specific message and a flag-specific response on an active link 17 within the LAG 15. Using these specific communications, network element 16 can quickly remove communications from failed link 17 without relying on timeouts or other mechanisms. According to a particular embodiment, the specific flag messages and responses use fields in legacy flag messages and responses, but provide additional information that can only be understood by properly enabled network elements 16 .

Specific embodiments for implementing the purge mechanism, extended flag protocol, and allocation adjustments are described in more detail below. However, although specific examples are provided in this specification, it should be understood that these examples are provided for illustrative purposes only, and thus system 10 may employ network elements 16 employing any suitable technology. Additionally, the particular embodiment shown and described with respect to system 10 is not intended to be exclusive or limiting. Although system 10 and elements within system 10 are described as having specific configurations and structures, it should be noted that these are logical descriptions and components and functions of system 10 may be logically and physically combined, separated, and assigned as appropriate. Furthermore, the functionality of system 10 and elements within system 10 may be provided by any suitable collection and arrangement of components.

FIG. 2 illustrates exemplary network elements 16 from system 10 of FIG. 1 . Network element 16 may include any suitable combination and configuration of components and modules. In the illustrated embodiment, network element 16 includes a LAG element 20 that facilitates forming LAG 15, a network element controller 21 that manages the operation of components within network element 16, and a port 22 for communicating over link 17, link 17 LAG 15 is formed by polymerization. LAG element 20 includes a LAG controller 24 and one or more LAG modules 26 . Each LAG module 26 includes: a media access control (MAC) client 28 ; and an aggregator 30 including a frame distributor 32 and a frame collector 34 . Port 22 includes one or more output queues 50 , central processing unit (CPU) queues 52 , and input queues 54 that facilitate communication of frames 56 . Generally, components within network element 16 facilitate communication between terminals 18 over network 12 . More specifically, components within network element 16 provide a clearing mechanism that facilitates moving sessions between links 17 in LAG 15 .

Network element controller 21 represents hardware including any suitable control logic capable of managing the operation of other components or modules within network element 16 . For example, network element controller 21 may be operable to load and execute software or other control logic from any suitable source.

Port 22 represents any suitable physical interface, including appropriate control logic, for interfacing with components in system 10 . In an embodiment, ports 22 represent physical interfaces between network elements 16 . A port 22 that transmits communications to other ports 22 may be referred to as a source port 22 . Alternatively, a port 22 that receives communications from other ports 22 may be referred to as a destination port 22 . Any port 22 can be used as source port 22 and destination port 22 when exchanging information in both directions during communication. Port 22 may include any suitable operational state. For example, port 22 may have a disabled state, a learning state, and a forwarding state. The learning state occurs when frames enter port 22, and the forwarding state occurs during normal traffic operation when frames enter and exit port 22. Network element 16 includes any suitable number of ports 22 . Each port 22 may have an associated physical address. For example, each port 22 may be assigned a unique globally administered MAC address. The ports 22 may be connected by a link 17 representing a communication channel between the ports 22 . Each port 22 may correspond to a link 17 . Communication between mobile terminals 18 may occur between links 17 within the LAG 15. Moving communication between links 17, for example, provides load balancing and maintains session availability if one or more links 17 in the LAG 15 fail.

Queues within port 22 also facilitate communication. Output queue 50 receives frames 56 from frame distributor 32 and holds frames 56 for transmission to network element 16 . According to a particular embodiment, output queue 50 transmits frames 56 on a first-in, first-out basis. Input queue 54 receives frames 56 and messages from network element 16 and provides frames 56 and messages to frame collector 34 . The CPU queue 52 provides messages and responses according to the flag protocol and the extended flag protocol to move sessions between links 17 within the LAG 15.

Element 20 represents any suitable combination of hardware and/or software that facilitates link aggregation. Element 20 includes a controller 24 and one or more modules 26 . Controller 24 represents hardware including any suitable control logic capable of managing the operation of other components or modules within LAG element 20 . For example, controller 24 facilitates the creation of LAGs 15, monitors the behavior of existing LAGs 15, and provides any suitable functionality to facilitate link aggregation. In a particular embodiment, controller 24 determines which links 17 can be aggregated, aggregates links 17, binds ports 22 to aggregator 30, and monitors LAG 15. In another embodiment, the network manager 14 manually controls the change of link aggregation.

Each negotiated LAG 15 has an associated module 26 which may be a logical description in an element 20. Module 26 facilitates the functionality of its associated LAG 15 and is used to implement changes in features within the LAG 15. For example, module 26 may be used to move communications occurring on link 17a within LAG 15 to link 17c when link 17 is active. As another example, if link 17a fails during communication, module 26 may be used to move communication on link 17a to link 17c.

Each module 26 includes a MAC client 28 and an aggregator 30 . MAC client 28 represents a logical media access controller for LAG 15, and aggregator 30 supports frame communication over link 17 and implements features within LAG 15. To support sending and receiving frames 56 between network elements 16 , an aggregator 30 is bound to one or more ports 22 .

While frames 56 are being sent and received by aggregator 30, the order of frames 56 is maintained during communication. Frame distributor 32 and frame collector 34 facilitate communication of frames 56 . Frame distributor 32 distributes frames 56 from terminals 18 on ports 22 using links 17 forming LAG 15. Frame distributor 32 ensures that frames 56 for a particular session are delivered to port 22 to prevent frames 56 from being out of order. Frame allocator 32 implements any suitable allocation algorithm that selects a link 17 for transmission of any given frame 56 or group of frames 56 belonging to a session. The chosen allocation algorithm prevents out-of-order frames 56 and duplication of frames 56 for the session. Frames for a given session are forwarded to port 22 based on the selected allocation algorithm. The allocation algorithm may be based on the destination address, the source address, a combination of the destination and source addresses, the address of the receiving port 22, or any other suitable criteria.

Frame collector 34 receives frames 56 from port 22 and transmits received frames 56 to terminal 18 . Frame 56 is forwarded out another port 22, which may be directly connected to terminal 18 or may be on the way to terminal 18, according to a particular embodiment. For example, frame collector 34 receives frame 56 from set of links 17 forming LAG 15. For any given port 22, the frame collector 34 transmits the frames 56 to the MAC client 28 in the order received from the port 22. Frame collector 34 may select frames 56 received from aggregation port 22 in any order. Because frame distributor 32 ensures that frames 56 maintain their order, frame collector 34 may not have to perform any reordering of frames 56 received from multiple links 17 while maintaining the frame order of the communication.

As noted above, network element 16 supports the Marker Protocol and the Extended Marker Protocol. Both protocols provide communication within aggregates. Using these protocols, for example, frame distributor 32 of network element 16a generates and distributes a token to frame collector 34 of network element 16b using a token protocol or an extended token protocol. Frame collector 34 of network element 16b distributes the marker response to frame distributor 32 of network element 16a using the marker protocol or the extended marker protocol. Messages in the Token Protocol including Tokens and Token Responses may have any suitable format. As mentioned above, the labeling protocol is used to move sessions between links 17 within the LAG 15. Using the flag protocol, the controller 24 generates a flag and transmits the flag on one or more active links 17 within the LAG 15. The frame collector 34 in the receiving network element 16 provides the flag response to the frame distributor 32 in the sending network element 16 . During removal of a session from one link 17, the session on the other link 17 can continue without interruption. Alternatively, network element 16 may use a labeling protocol to move sessions between two or more links 17 in an associated LAG 15. For example, controller 24 may generate and transmit flag messages over one or more links 17 and, upon receipt of responses, move sessions present on those links 17 .

The tokens and token responses (or specific tokens and specific token responses, respectively) in the extended token protocol may have any suitable format. In an exemplary embodiment, messages in the Extended Flags Protocol include the following format: Destination address = MAC address of the destination port Source address = MAC address of the source port that did not fail type=0x8809 subtype = flag-protocol version=0x01 TLV: Flag Information = 0x01, Flag Response = 0x02, Specific Flag = 0x03, Specific Flag Response = 0x04 Information length = 0x10 (1610) requester port requester system Requester Transaction ID = Fault Port for specific message Padding (pad) = 0x0000 terminator = 0x00 terminator length = 0x00 reserve frame check sequence

As mentioned above, the extended marker protocol can be used when a link 17 fails or becomes invalid, and the session is moved to one or more active links 17 . According to one embodiment of the extended flag protocol, frame distributor 32 in network element 16a provides specific flags on active links 17 within LAG 15. Frame collector 34 in network element 16b provides a specific flag response to frame distributor 32 . During removal of a session from one link 17, the session on the other link 17 can continue without interruption. Alternatively, network element 16 may use the extended label protocol to move sessions between two or more links 17 within the associated LAG 15. For example, controller 24 may generate and transmit flag messages over one or more links 17 and, upon receipt of responses, move sessions present on those links 17 .

According to one embodiment, an extended flag protocol is used to identify a failed or invalid link 17 in the message. For example, the message format may include the MAC address of the port 22 associated with the failed link 17 in the requester transaction ID field. As another example, the TLV field is used to identify the message as a specific flag or a specific flag response. Using the extended flag protocol, the frame distributor 32 generates specific flags that use the active links 17 within the LAG 15. Frame collector 34 sends a specific flag response to frame distributor 32, which determines that there are no ongoing frames 56 prior to moving the session. During the removal of a session from a failed or inactive link 17, communications on the other links 17 may continue without interruption.

Although the illustrated embodiment and the foregoing description focus on a specific embodiment of network element 16, system 10 may employ network element 16 having any suitable combination and configuration of components and modules that support the purge mechanism in LAG 15. Accordingly, the functionality performed by the specific elements shown may be separated or combined as appropriate, and some or all of these elements may be implemented by logic encoded in the medium. For example, the functions of frame distributor 32 and frame collector 34 may be suitably separated and/or combined, and either of their operations may be implemented by suitable control logic. Also, although shown as a single module, the functionality of some or all of the illustrated components of network element 16 may also be distributed among other elements of system 10 .

FIG. 3 is a flowchart 300 illustrating a method of moving sessions between ports 22 using a clearing mechanism using frame distributor 32 at sending network element 16 . The following description of the flowchart 300 is given with reference to the frame distributor 32 of the network element 16 as described above. However, any suitable element or combination of elements may implement the steps described below.

To facilitate communication between terminals 18 using LAG 15, at step 302 frame distributor 32 transmits frame 56 over link 17 in LAG 15. During the communication, at step 304, it is determined whether to move one or more sessions to another link 17 in the LAG 15. If the session is not moved, frame distributor 32 continues to distribute frames 56 over links 17 in LAG 15. On the other hand, if it is determined that the session is to be moved, then in step 306, the frame distributor 32 prohibits the distribution of the frame 56 through the link 17 in the LAG 15. Disabling allocation prevents, for example, additional frames 56 from being placed in output queue 50 . The frame distributor 32 enters a clear state and, at step 308 , discards the remaining frames 56 in the output queue 50 . During this clear state, the frame distributor 32 may drop all frames 56 intended to invalidate the link 17 , taking care of the dropped frames by means of upper layer recovery mechanisms. Thus, the CPU queue 52 can start sending the flag protocol message without waiting for the transmission of the remaining frames 56 in the output queue 50 . According to a particular embodiment, during the clear state, output queue 50 may continue to process and transmit control frames, such as bridge protocol data unit (BPDU) frames.

At step 310, the frame distributor 32 transmits a flag message to each link 17 in the LAG 15. For example, frame distributor 32 generates flag messages and places the flag messages in respective CPU queues 52 for transmission over link 17 . The flag message may include any suitable information for informing the network element 16 that traffic may be moved from a link 17 in the LAG 15 to another link 17 in the LAG 15. According to the example format of a message in the flag protocol provided above, if the value of the TLV is 0x01, the message is identified as a flag message.

At step 312, the frame distributor 32 receives the flag response. At step 314, it is determined whether there are outstanding additional flag responses. If the frame distributor 32 can receive additional responses, the method proceeds to step 316 where the frame distributor 32 waits to receive additional flag responses. From step 316, an additional flag response may be received at step 312, whereupon the method continues. In one embodiment, the frame distributor 32 waits to receive a flag response from each link 17 that has received a flag message. In this embodiment, the frame allocator 32 waits for different periods of time depending on the number of outstanding flag responses attached. In another embodiment, frame distributor 32 starts a timer while waiting to receive an additional flag response. Frame distributor 32 may use this timer to provide a configurable amount of time to wait for an additional flag response. The timer can be configured for any suitable period. Using this timer, the frame allocator 32 does not continue to wait for an additional flag response even if the additional flag response is not complete.

However, if additional flag responses are complete, then the method proceeds to step 318 . In step 318, the session is moved to another link 17 in the LAG 15. The frame allocator 32 returns to the non-clear state, and at step 320 enables the frame 56 to be allocated in the LAG 15. The session then continues on active link 17 in LAG 15.

The foregoing flowchart 300 illustrates exemplary operation of frame distributor 32 in network element 16 to move one or more sessions between ports 22 using a purge mechanism. However, the foregoing flowchart 300 and accompanying description merely illustrate exemplary methods of operation. Accordingly, various steps in flowchart 300 may be performed simultaneously and/or in a different order than shown. Additionally, frame allocator 32 may employ methods with additional steps, fewer steps, and/or different steps, so long as the method remains appropriate.

FIG. 4 is a flowchart 400 illustrating a method of responding to a decision to move a session between ports 22 through a clearing mechanism using frame collector 34 at receiving network element 16 . The following description of the flowchart 400 is given with reference to the frame collector 34 of the network element 16 as described above. However, any suitable element or combination of elements can carry out the actions described below.

To facilitate communication between terminals 18 using LAG 15, frame collector 34 receives frame 56 over link 17 at step 402. During the communication, at step 404, it is determined whether to move the communication to another link 17 in the LAG 15. If the session is not moved, frame collector 34 continues to receive frames 56 over link 17 in LAG 15. On the other hand, if it is determined that the session is to be moved, then at step 406 the frame collector 34 receives a flag message. For example, a decision is made to move the session to another link 17 in the LAG 15, and the frame distributor 32 sends a flag message to the frame collector 34 in the other network element 16. At step 408 , frame collector 34 determines the status of one or more output queues 50 corresponding to one or more input ports 22 . For example, controller 24 checks the status of output queue 50 using an interrupt or by reading the output queue status register. Additionally, frame collector 34 may determine the status of output queue 50 corresponding to each input port 22 .

At step 410 , frame collector 34 sends a flag response to frame distributor 32 . The flag response includes any suitable information in response to a flag message for mobile communication between links 17 . For example, the flag response confirms that there are no pending frames 56 prior to mobile communication. According to an example format of a message of the flag protocol, if the value of the TLV is 0x02, the message is recognized as a flag response.

As with flowchart 300, flowchart 400 and the accompanying description represent exemplary methods of operation only, and frame collector 34 and/or other suitable components may employ any suitable technique for moving communications between ports 22 using a clearing mechanism . Accordingly, various steps in flowchart 400 may be performed simultaneously and/or in a different order than shown. Additionally, frame collector 34 may employ methods with additional steps, fewer steps, and/or different steps, so long as the method remains appropriate.

FIG. 5 is a flowchart 500 illustrating a method for implementing an allocation adjustment mechanism in the LAG 15. The following description of the flowchart 500 is given by reference to the network element 16 as described above. However, any suitable element or combination of elements may implement the steps described below.

To facilitate communication between terminals 18 using LAG 15, allocation parameters are selected at step 502. A number of allocation parameters may be provided for use in determining how sessions are allocated among links 17 . Any suitable allocation parameters may be chosen. Allocation parameters include, for example, measures of link activity, configuration of system 10 , or status of network elements 16 . Any suitable element of system 10 may choose to assign parameters, such as network element 16 or network manager 14 may choose to assign parameters. Once the allocation parameters are selected, at step 504 an allocation function is selected. The allocation function allocates sessions among the links 17 according to selected allocation parameters. In one embodiment, each allocation parameter is associated with one or more allocation functions. In this embodiment, an allocation function is selected from among said associated allocation functions. Any allocation function associated with the selected allocation parameter can be selected. As with the allocation parameters, the allocation function may be selected by any suitable element of system 10 , such as network element 16 or network manager 14 . In step 506, network elements 16 exchange frames 56 over links 17 in LAG 15.

Selected allocation parameters and allocation functions can be adjusted during communication. During step 508 the performance of the LAG 15 is monitored. Monitoring performance includes monitoring any suitable parameter of system 10 , such as activity of link 15 or validity of frames 56 exchanged between ports 22 . For example, monitor parameters determined by selected allocation parameters. If a measure of link activity is selected as an allocation parameter, then during step 508 link activity is monitored. At step 510, it is determined whether the selected distribution function is to be adjusted. For example, network manager 14 may detect that sessions are not evenly distributed among links 17 within LAG 15. If it is determined not to adjust the selected allocation function, then continue monitoring the performance of the LAG 15 from step 508.

Alternatively, if it is determined that the selected allocation function is to be adjusted, network element 16 initiates a process for implementing an allocation adjustment mechanism. For example, the allocation function can be tuned if the performance of LAG 15 can be improved by using a different allocation function. As an example, if the measure of link activity is a selected allocation parameter, network element 16 may adjust the allocation function if link 17 is underutilized. In step 512, network element 16 refrains from allocating frame 56 to link 17 in LAG 15. Disabling allocation prevents, for example, additional frames 56 from being placed in output queue 50 . At step 514 , network element 16 discards the remaining frames 56 in output queue 50 . Thus, the CPU queue 52 can begin sending messages in the flag protocol without waiting for the transmission of the remaining frames 56 in the output queue 50 . Although frame 56 is dropped, output queue 50 may continue to process and output control frames, such as BPDU frames.

In step 516, the network element 16 sends a flag message to each link 17 in the LAG 15. For example, network element 16 generates a flag message, and CPU queue 52 sends the flag message. The flag message may include any suitable information for informing destination network element 16 that sessions may be reallocated between links 17 . According to an example format of a message of the flag protocol, if the value of the TLV is 0x01, the message is identified as a flag message.

At step 518, the network element 16 receives the flag response. At step 520 it is determined whether there are outstanding additional flag responses. If the network element 16 can receive an additional response, the method proceeds to step 522 where the network element 16 waits to receive an additional flag response. From step 522, an additional flag response may be received at step 518, whereupon the method continues. In an embodiment, the network element 16 waits to receive a flag response from each link 17 that received the flag message. In this embodiment, network element 16 waits for different periods of time depending on the number of outstanding flag responses attached. In another embodiment, network element 16 starts a timer while waiting to receive an additional flag response. Network element 16 may use this timer to provide a configurable period of waiting for an additional flag response. The timer can be configured for any suitable period. Using this timer, even if there is an outstanding additional flag response, when the timer expires, the network element 16 will not continue to wait for the additional flag response. In yet another embodiment, network element 16 may transmit a flag message over link 17 on which it has an outstanding flag response.

However, if there are no additional flag responses outstanding, then the method proceeds to step 524 . In step 524, the selected distribution function is adjusted. As mentioned above, tuning the selected allocation function to another allocation function associated with the allocation parameters can improve the performance of LAG 15. Using the flag protocol, the order of frames 56 within a session can be maintained even if the allocation function is adjusted. As noted above, any suitable component of system 10 may adjust the selected distribution function. For example, network manager 14 may adjust the allocation function automatically or through manual intervention. As another example, an automatic management tool detects the performance of LAG 15 and automatically adjusts LAG 15 by changing the selected allocation function. After adjusting the distribution between the links 17, the network element 16 makes it possible to distribute the frame 56 in the LAG 15 at step 526. Based on the adjusted allocation function, the session continues over link 17 in LAG 15.

The foregoing flowchart 500 illustrates exemplary operation of the network element 16 to implement the allocation adjustment mechanism in the LAG 15. However, the foregoing flowchart 500 and accompanying description illustrate only exemplary methods of operation. For example, network element 16 may adjust allocation parameters based on monitoring the performance of LAG 15. Adjustment of the allocation parameters also improves the performance of the LAG 15 by allocating sessions differently between the links 17. As another example, network element 16 adjusts allocation parameters and allocation functions to improve LAG 15 performance. Multiple steps in flowchart 500 may be performed simultaneously and/or in a different order than shown. Additionally, network element 16 may employ methods with additional steps, fewer steps, and/or different steps, so long as the method remains appropriate.

FIG. 6 is a flowchart 600 illustrating a method of implementing a specific flag message using the frame distributor 32 at the sending network element 16 . The following description of the flowchart 600 is given with reference to the frame distributor 32 of the network element 16 as described above. However, any suitable element or combination of elements may implement the steps described below.

To facilitate communication between terminals 18 using LAG 15, at step 602 frame distributor 32 transmits frame 56 over link 17 in LAG 15. At step 604, frame distributor 32 monitors link 17 for failure or other inefficiency. If link 17 has not failed, frame distributor 32 continues to distribute frames 56 over link 17 in LAG 15. On the other hand, if a particular link 17 does fail, then at step 606 the frame distributor 32 disables the source port 22 associated with the failed link 17 .

At step 608 , the frame distributor 32 generates a specific flag message on the active link 17 . This special flag message moves the session onto the active link using the extended flag protocol. At step 610, the frame distributor 32 transmits a specific flag message on the active link 17 in the LAG 15. For example, frame distributor 32 generates a special flag message, and CPU queue 52 sends the special flag message. The specific flag message may include any suitable information for informing network element 16 that a link 17 has failed and that communications will be moved to one or more active links 17 . According to an example format of a message of the extended flag protocol, if the value of the TLV is 0x03, the message is recognized as a specific flag message. Also, the above exemplary format also provides the MAC address of the faulty port 22 in the requester process ID field.

At step 612, frame distributor 32 receives a specific flag response. According to an example format of a message of the extended flag protocol, if the value of the TLV is 0x04, the message is recognized as a specific flag response. At step 614 the session is moved to one of the active links 17 . The frame distributor 32 makes it possible to distribute the frame 56 in the LAG 15 and continue the session through the link 17 in the LAG 15.

The foregoing flowchart 600 illustrates exemplary operations for implementing a specific flag message using the frame distributor 32 of the network element 16 . However, the foregoing flowchart 600 and accompanying description illustrate only exemplary methods of operation. For example, frame distributor 32 removes failed link 17 from LAG 15 when moving a session to a valid link 17 in LAG 15. Removing a failed link 17 is used to move a session to a valid link 17 while other sessions on other links 17 remain unchanged. As another example, the extended logo protocol may be used in conjunction with the logo protocol. In this example, sessions may be redistributed among active links 17 . Multiple steps in flowchart 600 may be performed simultaneously and/or in a different order than shown. Additionally, frame allocator 32 may employ methods with additional steps, fewer steps, and/or different steps, so long as the method remains appropriate.

FIG. 7 is a flowchart 700 illustrating a method of responding to a specific flag message using the frame collector 34 at the receiving network element 16 . The following description of flowchart 700 is given with reference to frame collector 34 of network element 16 as described above. However, any suitable element or combination of elements may implement the steps described below.

To facilitate communication between terminals 18 using LAG 15, at step 702 frame collector 34 receives frame 56 over link 17 in LAG 15. At step 704, the frame collector 34 monitors for certain flag messages. For example, in conjunction with normal processing of management messages received on link 17, frame collector 34 may detect specific flag messages sent from frame distributor 32 of remote network element 16 to notify framers of a failed or invalid link 17. Collector 34. If no specific flag message is received at step 706 , frame collector 34 continues to monitor for specific flag messages at step 704 . If a specific flag message is received, the method continues to step 708 .

At step 708 , frame collector 34 determines the status of one or more output queues 50 corresponding to port 22 associated with failed link 17 . For example, controller 24 checks the status of output queue 50 using an interrupt or by reading the output queue status register. At step 710 , frame collector 34 sends a specific flag response to frame distributor 32 at sending network element 16 . The specific flag response includes any suitable information for responding to specific flag messages for mobile communications between links 17 following a link 17 failure. For example, the specific flag response confirms that there are no pending frames 56 prior to mobile communication. According to an example format of a message of the extended flag protocol, if the value of the TLV is 0x04, the message is recognized as a flag response. Furthermore, the exemplary format described above also provides the MAC address of the failed port 22 in the requester transaction ID field.

The foregoing flowchart 700 and accompanying description represent only exemplary methods of operation, and frame collector 34 and/or other suitable components may respond to LAG-specific flag messages using any suitable technique. Accordingly, various steps in flowchart 700 may be performed simultaneously and/or in a different order than shown. Additionally, frame collector 34 may employ methods with additional steps, fewer steps, and/or different steps, so long as the method remains appropriate.

While the invention has been described in several embodiments, variations and modifications will be suggested to those skilled in the art, and the invention is intended to cover such variations and modifications as fall within the scope of the appended claims of the invention.

related application

This application claims priority to U.S. Provisional Application Serial No. 60/670,369, filed April 12, 2005, entitled "Link Aggregation and Network Management Techniques," which is incorporated by reference into this article.

Claims (21)

1. A method for redistributing sessions between links in a link aggregation group, comprising the steps of: aggregating a plurality of physical links connected between the first network element and the second network element into a link aggregation group; distributing received frames among output queues corresponding to respective physical links, the frames being associated with one or more sessions; deciding to move at least one of said sessions between said physical links; inhibiting allocation of additional received frames to output queues corresponding to at least one of said physical links; transmitting a flag message on the at least one physical link; clearing frames from an output queue corresponding to the at least one physical link; receive flag responses; and Responsive to receipt of the flag response such that subsequently received frames may be distributed among the physical links. 2. The method of claim 1, further comprising the steps of: Transmit flag messages on each physical link; Disallow frame allocation to all physical links; clear the output queues associated with all physical links; and A flag response is received on each physical link before allowing subsequently received frames to be distributed among the physical links. 3. The method of claim 1, wherein distributing frames among output queues comprises distributing frames among output queues according to a distribution algorithm, further comprising: adjust the allocation algorithm; and This allows subsequently received frames to be distributed among the physical links according to an adapted distribution algorithm. 4. The method of claim 1, wherein transmitting a marker message on the at least one physical link comprises transmitting a marker message using an administrative queue associated with the at least one physical link. 5. The method of claim 1, further comprising the step of: receiving a remotely transmitted flag message at a port corresponding to one of the physical links; monitor the queue corresponding to the port; and In response to the monitoring of the queue corresponding to the port, a flag response is transmitted on the physical link corresponding to the port. 6. The method of claim 5, further comprising transmitting the flag response on the physical link corresponding to the port in response to all frames in the queue corresponding to the port having been processed. 7. The method of claim 1, wherein allocating additional received frames to and purging frames from an output queue corresponding to the at least one physical link is inhibited, Frames intended for the at least one physical link result in being dropped such that the dropped frames have to be retransmitted. 8. A network element comprising: multiple ports; an aggregator associated with two or more of the ports aggregated to form a link aggregation group, the aggregator comprising a frame distributor and a frame collector, wherein the frame collector is operable to receive Inbound frames received on aggregated ports and the frame distributor is operable to distribute outbound frames associated with one or more sessions among the output queues corresponding to each aggregated port ; a controller configured to decide to move one or more sessions within the link aggregation group, wherein the controller is further configured to disable allocation of additional received frames to output queues corresponding to at least one aggregated port; and Wherein the frame distributor is also used to send a flag message on the at least one aggregation port to clear frames from the output queue corresponding to the at least one aggregation port, and the frame collector is also used to receive a flag response , and in response to receipt of a Flag Response, the frame distributor is further operable to enable distribution of subsequently received frames among the aggregation ports. 9. The network element of claim 8, wherein: The frame allocator is also operable to transmit a flag message on each physical link, disable frame allocation to all physical links, and clear output queues associated with all physical links; and The frame collector is further operable to receive flag responses on the respective physical links before the frame distributor enables distribution of subsequently received frames among the physical links. 10. The network element according to claim 8, wherein the frame distributor is further configured to distribute outbound frames among the output queues corresponding to each aggregation port according to a distribution algorithm, the distribution algorithm is adjusted, and it is possible to Subsequent received frames are distributed among the aggregated ports according to the adjusted distribution algorithm. 11. The network element of claim 8, wherein the frame collector is further operable to: receiving a remotely transmitted flag message at one of said aggregation ports; monitoring the queue corresponding to the aggregation port that received the remotely transmitted flag message; and In response to monitoring the queue, a flag response is sent on the aggregation port that received the remotely transmitted flag message. 12. The network element of claim 11, further comprising sending the flag response on the aggregate port that received the remotely transmitted flag message in response to all frames in the queue corresponding to the aggregate port having been processed. 13. The network element of claim 8, wherein allocating additional received frames to and purging frames from the output queue corresponding to the at least one physical link is inhibited, causing frames intended for the at least one physical link to be dropped such that the dropped frames have to be retransmitted; receiving a remotely transmitted flag message on one of said physical links; determining the state of the queue corresponding to the physical link transmitting the remotely transmitted flag message; and A flag response is transmitted over the physical link that transmitted the remotely transmitted flag message. 14. A logic for redistributing sessions among links in a link aggregation group, the logic encoded on media and operable to perform the steps of: aggregating a plurality of physical links connected between the first network element and the second network element into a link aggregation group; distributing received frames among output queues corresponding to respective physical links, the frames being associated with one or more sessions; deciding to move at least one of said sessions between said physical links; prohibiting allocation of additional received frames to output queues corresponding to at least one physical link; transmitting a flag message on the at least one physical link; clearing frames from an output queue corresponding to the at least one physical link; receive flag responses; and Responsive to receipt of said flag response enables distribution of subsequently received frames among the physical links. 15. The logic of claim 14, when executed, further operable to: Transmit flag messages on each physical link; Disallow frame allocation to all physical links; clear the output queues associated with all physical links; and A flag response is received on each physical link before allowing subsequently received frames to be distributed among the physical links. 16. The logic of claim 14, wherein allocating frames among output queues comprises allocating frames between output queues according to an allocation algorithm, the logic, when executed, being further operable to perform the steps of: adjust the allocation algorithm; and This allows subsequently received frames to be distributed among the physical links according to an adapted distribution algorithm. 17. The logic of claim 14, wherein transmitting a marker message on the at least one physical link comprises transmitting a marker message using an administrative queue associated with the at least one physical link. 18. The logic of claim 14, when executed, further operable to: receiving a remotely transmitted flag message at a port corresponding to one of the physical links; monitor the queue corresponding to the port; and In response to the monitoring of the queue corresponding to the port, a flag response is transmitted on the physical link corresponding to the port. 19. The logic of claim 18, when executed, further operable to transmit on the physical link corresponding to the port in response to all frames in the queue corresponding to the port having been processed The flags respond. 20. The logic of claim 14, wherein allocating additional received frames to and purging frames from an output queue corresponding to at least one physical link is inhibited, resulting in Frames intended for the at least one physical link are dropped such that the dropped frames have to be retransmitted. 21. A network element comprising: means for aggregating a plurality of physical links connected between the first network element and the second network element into a link aggregation group; means for distributing received frames among output queues corresponding to respective physical links, the frames being associated with one or more sessions; means for determining to move at least one of said sessions between said physical links; means for inhibiting allocation of additional received frames to an output queue corresponding to the at least one physical link; means for transmitting a flag message on said at least one physical link; means for clearing frames from an output queue corresponding to the at least one physical link; and Means for receiving a flag response, and in response, enabling a subsequently received frame to be distributed among the physical links.

Images