Understanding Link Aggregation, Trunk Technologies, and Bonding Modes in Network Switches

In most scenarios the terms "link aggregation" and "trunk" are used interchangeably, but they have distinct meanings. Link aggregation combines multiple physical network links in parallel to increase bandwidth and provide redundancy, creating a single logical link that avoids loops in spanning‑tree protocols. Vendors use different names (e.g., Cisco EtherChannel, Brocade LAG) but the IEEE 802.1ax standard (formerly 802.3ad) defines LACP, which dynamically configures aggregation without vendor lock‑in.

When multiple ports are bundled between switches, the Spanning Tree Protocol (STP) would normally see them as a loop; link aggregation solves this by presenting the bundle as one logical link, preventing loops.

Trunk technology is typically used between switches to increase reliability through redundant links, expand port count via cascading, and improve link utilization. Several trunk scenarios exist:

ISL Trunk

ISL (Inter‑Switch Link) trunking connects switches to add more ports, enable cascading, and support long‑distance transmission (e.g., Fibre Channel switches using single‑mode fiber).

VSAN Trunk

This is a Cisco Fibre Channel concept where multiple VSANs share a single ISL link, similar to VLANs on IP switches, providing layer‑2 isolation. Brocade calls a comparable feature "Zoning".

VLAN/Ethernet Trunk

VLAN trunking tags frames so that hosts in the same VLAN on different switches can communicate. The trunk ports on the cascading links carry traffic for multiple VLANs, increasing link utilization without needing a separate ISL for each VLAN. Standard 802.1Q and Cisco's proprietary ISL protocol are commonly used (ISL works only on Cisco switches).

VLAN/Ethernet Trunk Summary

1. VLANs span multiple switches, each switch may have the same VLAN IDs. 2. A single physical link (the trunk) carries traffic for all those VLANs. 3. Only trunk links, not access links, can carry multiple VLANs.

Link Bonding / Port Bonding

Servers or storage devices can use bonding to achieve redundancy and load balancing. Linux kernel supports seven bonding modes; some (e.g., LACP) work best when paired with a switch that also supports LACP.

Balance‑rr (Round‑robin) mode

All active interfaces are placed in a circular queue; packets are transmitted sequentially across the interfaces, providing even distribution.

Active‑backup (AP) mode

One interface is active while the others are standby; if the active link fails, a standby takes over.

Balance‑XOR mode

Packets are sent based on a hash of source/destination MAC addresses (or a user‑defined policy), ensuring that packets of the same flow follow the same interface.

Broadcast mode

Every packet is transmitted on all interfaces, useful when two independent networks must receive the same traffic.

802.3ad (LACP) mode

Implements the IEEE LACP dynamic aggregation protocol; recommended when the opposite switch supports LACP.

Balance‑tlb (Transmit Load Balancing) mode

Distributes outgoing traffic based on each slave's current load and processing capability, while incoming traffic follows the default route.

Balance‑alb (Adaptive Load Balancing) mode

Extends TLB by also balancing incoming IPv4 traffic, using ARP negotiation to distribute receive load across slaves.

Warm tip: Scan the QR code below or search for “ICT_Architect” to follow the public account and obtain more ICT‑related resources.

Click the “Original Link” to visit the “ICT Architect eBook Library” for additional electronic materials.