MC-LAG

LAG is a method of inverse multiplexing on several channels of local area networks, increasing throughput and redundancy. This is defined by the IEEE 802.1 AX-2008 standard, which describes the following: “link aggregation allows one or more links to be combined together to form a group of channels so that the MAC client can process it as if it were a single connection.” ^{[ 1]} The channel level of access is achieved by using the LAG together with one MAC address for all device ports in the group. LAG can be configured both statically and dynamically. Dynamic Latency uses a peer-to-peer control protocol called the Link Aggregation Control Protocol (LACP). This LACP protocol is also defined in the 802.1 AX-2008 standard.

LAG can be implemented in two ways. LAG N and LAG N + N. LAG N is a load balancing mode between LAG and LAG N + N that provides the user with a sense of expectation.

The LAG N protocol dynamically distributes the traffic load across the working channels within the LAG. And thus maximizes the use of the group, providing improved stability and bandwidth, in case there are jumps in the speed of Ethernet-connection.

For another type of resilience between 2 nodes, the full implementation of LACP supports separate working / standby LAGs. For LAG N + N, work relationships as a group will switch to backup if one or more or all of the links in the work group fail. Note: LACP marks links that are in standby mode with the “no synchronization” flag.

The MC-LAG adds node level redundancy to the normal link level redundancy that the LAG provides. This allows two or more nodes to share a common LAG endpoint. Multiple nodes represent a single logical LAG to the remote end. Note that MC-LAG implementations are vendor-specific, but interconnecting switches remain externally compliant with the IEEE 802.1 AX-2008 standard. ^[2] The nodes in the MC-LAG cluster interact to synchronize and negotiate automatic failover (failover). Some implementations may support administrator-initiated (manual) failovers.

The diagram here shows four configurations:

In this illustration, you can see a comparison of the LAG and MLAG functionalities.

1. Switches A and B are each configured to group four discrete channels (as indicated in green) into a single logical connection with four times the bandwidth. The standard LACP protocol ensures that if any of the links goes down, traffic will be distributed among the other three.
2. Switch A is replaced by two switches A1 and A2. They communicate with each other using their own protocol, and thus can be disguised as a single “virtual” switch that runs a common LACP instance. Switch B does not know that this is associated with multiple groups.
3. Switch B is also replaced by two: B1 and B2. If these switches are from a different vendor, they can use a different proprietary protocol between themselves. But the “virtual” switches A and B still communicate using LACP.
4. The intersection of two bonds to form X does not logically matter, nor does the intersection of links in a normal LAG. However, physically it provides significantly improved fault tolerance (high reliability). If either switch fails, LACP restructures the path in just a few seconds. Work continues with paths existing between all sources and destinations, albeit with reduced bandwidth.

HA configuration is more advanced than spanning tree. The load can be shared between all links during normal operation, while the spanning tree should disable some links to prevent loops.

• IEEE 802.1aq ( Shortest Path Bridging )
• Lacp
• Link aggregation