Hands-On Lab: Configuring Layer 3 EtherChannel

This hands-on lab configures a routed EtherChannel between two Layer 3 switches, bundling two physical links into a single logical routed interface rather than a switched trunk, and verifies OSPF forms a single neighbor relationship across the bundle rather than one per physical link.

Layer 3 EtherChannelRouted Port-ChannelNo Switchport Configuration

~3 min read · Updated Sep 23, 2026

Lab Objective

Configure a Layer 3 EtherChannel between two Layer 3-capable switches, assigning an IP address directly to the port-channel interface rather than to individual physical interfaces, and verify OSPF forms exactly one neighbor relationship across the bundled link rather than treating each physical member as a separate path.

Lab Purpose

Every EtherChannel configured earlier in this series was a Layer 2 switched bundle. A Layer 3 EtherChannel instead combines multiple physical links into a single routed interface, useful for high-bandwidth links between distribution or core Layer 3 switches where the connection needs an IP address directly rather than functioning as a VLAN trunk.

Lab Topology

Switch1 (Layer 3 capable) ---- Gi1/0/1, Gi1/0/2 ----
                                                  Switch2 (Layer 3 capable)

Po1 on Switch1: 10.99.1.1/30
Po1 on Switch2: 10.99.1.2/30

Task 1: Configure the Physical Interfaces as Layer 3 Channel Members

Remove any switchport configuration from Gi1/0/1 and Gi1/0/2 on both switches, and bundle them into a Layer 3 port-channel.

Task 2: Assign an IP Address to the Port-Channel Interface

Configure the IP address directly on Po1 on both switches, rather than on any individual physical interface.

Task 3: Verify the Bundle Formed Correctly

Confirm both member links are bundled and active.

Task 4: Verify Basic Connectivity

Ping between the two Po1 addresses.

Task 5: Enable OSPF and Verify a Single Neighbor Relationship

Enable OSPF on the Po1 interfaces and confirm exactly one neighbor forms, not two.

Solution and Verification

Switch1(config)# interface range gigabitethernet1/0/1-2
Switch1(config-if-range)# no switchport
Switch1(config-if-range)# channel-group 1 mode active

-- "no switchport" is what makes this a
-- Layer 3 bundle rather than the Layer 2
-- switched bundles configured in earlier
-- EtherChannel labs

Switch2(config)# interface range gigabitethernet1/0/1-2
Switch2(config-if-range)# no switchport
Switch2(config-if-range)# channel-group 1 mode active

Switch1(config)# interface port-channel 1
Switch1(config-if)# ip address 10.99.1.1 255.255.255.252
Switch1(config-if)# no shutdown

Switch2(config)# interface port-channel 1
Switch2(config-if)# ip address 10.99.1.2 255.255.255.252
Switch2(config-if)# no shutdown

Switch1# show etherchannel summary

Group  Port-channel  Protocol   Ports
------+-------------+-----------+-----------------------
1      Po1(RU)       LACP       Gi1/0/1(P)  Gi1/0/2(P)
-- "R" in the flags indicates this is a
-- Layer 3 (routed) port-channel, distinct
-- from the "S" (switched/Layer 2) flag
-- seen in earlier EtherChannel labs

Switch1# ping 10.99.1.2

!!!!!
Success rate is 100 percent (5/5)

Switch1(config)# router ospf 1
Switch1(config-router)# network 10.99.1.0 0.0.0.3 area 0

Switch2(config)# router ospf 1
Switch2(config-router)# network 10.99.1.0 0.0.0.3 area 0

Switch1# show ip ospf neighbor

Neighbor ID     Pri   State           Address         Interface
2.2.2.2          1     FULL/  -        10.99.1.2        Port-channel1
-- Exactly one neighbor relationship formed
-- over the single logical Port-channel1
-- interface, rather than OSPF attempting to
-- form separate adjacencies over Gi1/0/1
-- and Gi1/0/2 individually

Key Takeaway

The no switchport command is the single defining difference that converts an EtherChannel from a Layer 2 switched bundle into a Layer 3 routed one — once configured this way, the IP address and any routing protocol configuration belongs entirely on the logical Port-channel interface, never on the individual physical members, exactly mirroring how VLAN and trunk settings were applied to the logical bundle rather than individual members in the Layer 2 EtherChannel labs covered earlier in this series.

Written & researched by Dr. Shahin Siami

Related Articles

Hands-On Lab: Configuring iBGP and the Full-Mesh Requirement

This hands-on lab configures iBGP among three routers within a single autonomous system, demonstrating the full-mesh peering requirement by deliberately omitting one peering relationship and observing the resulting route propagation failure.

Continue

Hands-On Lab: Configuring eBGP Between Two Autonomous Systems (CCNP Depth)

This hands-on lab configures eBGP between two routers in different autonomous systems with a full verification workflow, examining the BGP table, AS-path attribute, and confirming routes are correctly installed with eBGP's default administrative distance.

Continue

Hands-On Lab: Configuring Policy-Based Routing (PBR)

This hands-on lab configures Policy-Based Routing on a router to send traffic from a specific source subnet out a different path than the normal routing table would select, overriding the destination-based forwarding decision that every previous routing lab in this series relied on.

Continue

Hands-On Lab: Configuring Route Leaking Between VRFs

This hands-on lab configures selective route leaking between two VRFs using route targets and a shared services VRF, allowing specific routes to cross the otherwise strict isolation boundary established in the previous VRF-Lite lab.

Continue

Hands-On Lab: Configuring VRF-Lite

This hands-on lab configures VRF-Lite on a router to maintain two completely separate routing tables for two different customer networks sharing the same physical router, verifying each VRF's traffic remains isolated despite using overlapping IP address space.

Continue

Hands-On Lab: Configuring LACP Fast Rate

This hands-on lab configures LACPDU fast rate on an EtherChannel bundle, reducing the interval between LACP control packets to accelerate detection of a failed member link compared to the default slow rate.

Continue