Link Aggregation and Balancing Methods

Question

This might be a silly question, but I am not really sure about how this works.

When you form Link Aggregation (or EtherChannel in Cisco terms), the traffic going through that logical port will be load-balanced based on the device setting.

For example, in Cisco switches the default will be souｒce mac address based. And you can change that setting in a way that the traffic will be distributed almost equally.

So far, it's fine.

But how about when the load-balance method is src ip and dst ip but the switch is NOT the server's/client's gateway?

For example, if your NW is something like below,

Internet--FW--SW--SW--Servers/Clients

and the Link Aggregation is formed between the SWs but the default gateway is FW.

In other words, all SWs do is not routing but switching.

What I'm wondering is whether ip-based load-balancing would properly work if the SWs are not doing routing because Layer 2 switching doesn't really require that devices read IP header as well.

Does anybody know what would happen if ip-based load-balacing is configured in the above environment?

Answer 1

Many Cisco switches can look at higher layer data, including IP addresses, under certain circumstances, and this is one of them. Other circumstances are for things like DHCP snooping, ACLs (e.g. QoS marking based on IP source/destination addresses and/or source/destination TCP or UDP ports), IGMP snooping, IP Source Guard, etc.

Other vendors may, or may not, have similar switch features.