LACP overview

Last Updated : Oct 30, 2026 |
Prolog information
Metadata
Keywords:
LACP periodically transmits keepalive frames over an aggregated link to verify the connectivity of each member link and to trigger a failover in the event of a malfunction. LACP can be configured with a fast or slow periodic rate, resulting in an effective detection time of approximately ~3 seconds or ~90 seconds, respectively, assuming the standard default timeout of three missed LACP PDUs.
In VoIP environments, the fast rate is recommended to ensure rapid failover and minimize call disruption, while the slow rate can significantly delay failure detection and impact active calls.

LACP Load Balancing Hash Policy

For KVM workloads carrying real-time VoIP traffic, Layer 3+4 hashing is recommended to ensure even traffic distribution, minimize jitter, and avoid congestion on individual links. However, the selected policy must also be supported and configured consistently on the customer switch; mismatched policies can lead to uneven traffic flows and degraded performance.
  • Layer 2 (MAC-based hashing): Uses source and destination MAC addresses to determine the outgoing interface. This provides predictable distribution but may not effectively balance traffic when most flows share the same MAC pairs, which is common in virtualized servers.
  • Layer 3+4 (IP and port-based hashing): Uses source/destination IP addresses and transport-layer ports (TCP/UDP) in the calculation. This method provides more granular distribution across links and is generally preferred for environments running multiple sessions or real-time applications, such as VoIP. It reduces the risk of oversubscription on a single link and helps maintain consistent call quality.
Related links
Adding a bond using the network configuration script