How does LACP Fast-Switchover work?

[Mitrixsen]

Hello, everyone. I have some questions regarding LACP F-Switchover.

Here is my topology:

From my understanding, LACP Fast-Switchover can be used in situations where we have a 1:1 redundancy. So we basically have one port in the EC (Etherchannel) forwarding data while the other one is in hot standby mode ready to take over if the main port fails.

Without LACP F-Switchover, if the active port goes down, the PO itself will also go down and then switch to the hot standby port?

But with Fast Switchover, the PO transitions the G0/1 port instantly into active and doesn't go down?

I have two questions regarding this:

1. Why exactly does the EC also go down when the active port goes down and hot-standby port takes over? What is happening in the background here? Why can't it just stay up as with LACP F-Switchover?

2. Are there any cases where we would even use hot standby ports with LACP? Isn't it more logical to have all the links forwarding and have EC load-balance between them?

Thank you.

David

[M02@rt37]

Hello @Mitrixsen 

When an active port in an ethercannel fails without LACP Fast switchover, the entire Po can momentarily go down because LACP must renegotiate the channel with the remaining links. This happens because LACP needs to reconfigure the active paths and ensure consistency in the network, which causes a short disruption. During this time, the Port-Channel doesn’t have any active links until the failover is complete, leading to the channel appearing down briefly. LACP Fast switchover, however, pre-configures a standby port to take over immediately when the active port fails, allowing the Port-Channel to remain "up" without needing to renegotiate, thus preventing any noticeable downtime.

While using hot standby ports in LACP setups isn't common, it can be useful in scenarios where network stability is crucial, and immediate failover is required without relying on load balancing. For instance, in environments with critical links, a hot standby port ensures that traffic is instantly rerouted in case of failure, maintaining uninterrupted service. However, in most cases, it is more efficient to use all links in an EtherChannel for load balancing (?) as this maximizes available bandwidth and provides redundancy by allowing any link to take over if another fails, without needing a dedicated standby link.

[Mitrixsen]

Hello.

I've often seen this recommended in scenarios where you have a max-bundle set to 1 link so 1 link is active and one is standby. Would this cause any problems if there were more links in the bundle and F-Switchover was configured?

David

[MHM Cisco World]

it  not problem but it not usual

All engineer use at least two link in PO (one link have no sense to use) 

So config max two 2 and add new port "H" this will solve issue of PO down (PO i use for etherchannel' other vendor use term EC as you)

Without  fast 

One of two port down' the PO not down' since it still have one P port'  and H will become  P port  without any drop in traffic 

With fast 

Same as without but faster 

MHM

[M02@rt37]

@Mitrixsen 

No, configuring LACP Fast-Switchover with more links in the bundle and max-bundle set to 1 will not cause technical problems, but it will limit the utilization of available bandwidth. In this setup, only one link is active at a time, while the others remain in standby. If the active link fails, Fast-Switchover ensures an immediate transition to one of the standby links, minimizing downtime. However, even with multiple links in the bundle, only one will ever be active due to the max-bundle setting, leading to underutilization of the additional links. This configuration is effective for redundancy and quick failover but does not take full advantage of the potential bandwidth that could be achieved by using more links simultaneously. If higher bandwidth is needed, it would be more efficient to allow multiple active links by increasing the `max-bundle` value, which enables load balancing across the links while still providing redundancy.

[Flavio Miranda]

@Mitrixsen 

The port-channel should not go down if one link became unavailable. And this is independent of fast switchover. Hoever, you need to configure lacp max-bundle first.

https://www.cisco.com/c/en/us/td/docs/ios-xml/ios/cether/command/ce-xe-3se-3850-cr-book/ce-xe-3se-3850-cr-book_chapter_00.html#wp9924202450

[MHM Cisco World]

you config max bundle 1 
that why one port is "P" other is "H" am I correct ?

MHM

[Joseph W. Doherty]

#1 Answered in M02@rt37 reply, I believe.  If you don't believe so too, let us know.

#2 Usage cases?  Sure, several come to my mine, but don't see them being commonly encountered.

As to EC being logically better using both links activity, depends on what all your considerations comprise.

Let's consider an unusual example case.

You're working with two different MAN/WAN SPs.  One or both charge based on CIR and/or data usage.  Depending on the costing model, it's possible you only want to use the one link as a standby.  Ideally you want it to switch to active as quickly as possible.

I've dealt with such costing issues at L3, but if you were only doing L2, what are your options?  Perhaps STP costing, but might a standby EC link be better?  (Especially at certain times of the day, or some other temporary bandwidth need change, where the standby is reconfigured to be fully active?)

Anytime $ is a consideration, it may have a huge design impact.  ($ has its own logic.)

[Mitrixsen]

That's a great explanation, I appreciate your help here. I just wanted to know one extra thing regarding the first question. Would this cause any problems if there were more links in the bundle and F-Switchover was configured? Since I've only ever seen this with a maximum of 2 links, as this post implies.

David

[Joseph W. Doherty]

With more than 2 links, don't see fast switch as a problem.

Possibly a good usage case, you've hit the maximum active, 8, and you have 1 or more spares that you want to replace a failed link as quickly as possible.