Will there be any problems if the algorithms on both ends of a switch link aggregation are inconsistent?

What is Link Aggregation?

Link aggregation is a technology that bundles multiple physical links into one logical link to increase link bandwidth and redundancy.

Depending on whether the Link Aggregation Control Protocol LACP is enabled, link aggregation is divided into two aggregation modes: "static" and "dynamic (LACP)".

• Static aggregation: The establishment of Eth-Trunk and the addition of member interfaces are manually configured without the involvement of link aggregation control protocol. All activities participate in traffic sharing, which is also called manual load balancing mode.
• Dynamic aggregation: Based on IEEE802.3ad standard LACP, Link Aggregation Control Protocol is a protocol that implements dynamic link aggregation and deaggregation. LACP exchanges information with the peer end through the Link Aggregation Control Protocol Data Unit LACPDU.

Today I will talk about "static aggregation". The aggregation algorithm is indispensable for configuring static aggregation. There are mainly: based on destination IP, based on destination MAC, based on source destination IP, based on source destination MAC, based on source IP, based on source MAC.

Then there is a classic question: Will there be any problem if the algorithms at both ends of the switch link aggregation are inconsistent? As shown in the following figure:

Problem Analysis

Let me tell you the answer first:

Let's take a brief look at the algorithm logic: the switch will hash the MAC and IP address fields in the data frame into a HASH-KEY value, and then find the physical outbound interface in the aggregation forwarding table to send it out. To ensure that the same data stream is forwarded on the same physical link, why must it be consistent? To prevent the address table of the other end from drifting.

Therefore, the switch's aggregation algorithm only determines TX (transmit) and has nothing to do with RX (receive), as shown below:

When PC1 and PC2 communicate, SW1 can forward PC1's data to SW2 from different outbound interfaces based on different algorithms;

SW2 doesn't care whether you come from port 1 or port 2, it only cares whether the data flow from PC1 to PC2 is continuously maintained on the same physical link. Otherwise, if PC1 comes from port 1 and port 2, wouldn't the address table of SW2 be disordered? The direct result of disorder is communication abnormality;

Similarly, when SW2 transfers the data from PC2 to SW1, SW1 will not care which port you give it to me. It only cares about "whether you can always give it to me through this physical port".

Therefore, under the standard convergence algorithm of the switch, inconsistent algorithms at both ends will not affect communication.