Link Aggregation vs LACP – What is the difference?

One of our incredibly talented Instructors, Tony Bourke, has put together a short video, The Confusion Between Link Aggregation and LACP, explaining in detail the differences between LACP and Link Aggregation.

So, what is Link Aggregation? What are we talking about here?

Simply put, it’s a way to bundle a group of individual links together in order for them to act as a single logical connection. Some benefits or advantages of link aggregation include: additional bandwidth between connected devices, provide fast and transparent recovery in case one or more of the individual links fails (redundancy) which increases resilience, and load balancing of the connections which can boost capacity. All this can be particularly important in congested and loaded networks.

At this stage I think it’s important to explain or review some of the confusing terminology used within discussions of link aggregation.

• Link aggregation group (LAG) is a group of ports combined together. Arista also refers to this as a port channel. 
• LAGs can be linked together into a network switch which creates a link aggregation group.
• Importantly, the active monitoring protocol allowing connected devices to either remove or include specific links from the LAG is called Link Aggregation Control Protocol (LACP).
  

Well then, what is Link Aggregation Control Protocol? Why is it used?

First, it’s good to understand that LACP is optional and will make a LAG dynamic or static, whereas Link Aggregation is a standard or an agreed upon set of behaviors. In Tony’s video, he describes LACP as a sort of ‘mis-cabling protocol’ and it’s used, generally speaking, to ensure the devices are all connected to the correct devices.

Here is a visual of what happens when LACP is not used.

There is no way of determining whether or not all links are actually connected to the correct place.

In this instance, the 4th link is not connected to the correct device and ¼ of the traffic ends up ‘black holed’.

Now we have a visual of what happens when LACP is used.

As you can see, link 4 is identified as not receiving a system link and is therefore removed from the LAG.

If we look at the show commands in the image below, we see that the three links are bundled together with the same system ID, however no LACP PDUs are being received so it has dynamically removed it from the LAG.

This is pretty much what LACP is used for – to ensure no traffic is sent if something is not connected to the correct device. 

Further Reading

If you want to delve into this topic even further we recommend looking into Arista’s ACE:L2 course.

You can also head over to our YouTube channel where we have more technical networking videos. Click below to start watching.

Find out more about Tony Bourke from his blog https://datacenteroverlords.com/