WAN Smoothing vs. Forward Error Correction (FEC): Which to Use

  • Updated

WAN Smoothing and Forward Error Correction (FEC) are two SpeedFusion features that both fight packet loss and jitter, and people often aren't sure which to turn on. They solve the problem in different ways, with different bandwidth costs and different ideal environments. Here's how to choose.

Forward Error Correction (FEC)

FEC adds redundant "parity" data alongside your real traffic. If some packets are lost in transit, the far end can rebuild them from the parity data without asking for a retransmission. That's important when retransmissions are slow or expensive.

  • Overhead is moderate and adjustable (low/medium/high).
  • Highly effective on high-latency or extremely lossy environments, such as LEO satellite (Starlink) or remote cellular networks, where round-trip times are high and retransmitting a lost packet is costly.
  • Ideal for video streaming and multicast applications, where you want to recover lost packets cleanly without the delay of a resend.

WAN Smoothing

WAN Smoothing takes a more aggressive approach: it duplicates your packets across multiple links (or sends multiple copies) so that if one link drops packets, an identical copy still arrives. The receiver discards the duplicates. This produces the smoothest possible real-time experience, at the cost of multiplying the bandwidth used.

  • Best suited for ultra-low-latency, real-time applications like VoIP, video conferencing, and broadcast. Peplink's own guidance is that WAN Smoothing is only needed when the environment has a problem with VoIP or video streaming (packet drop, jitter).
  • Because of its high bandwidth consumption, it is typically used for critical sessions, not for bulk downloads.

How much bandwidth does WAN Smoothing use?

WAN Smoothing consumes bandwidth as a multiple of your original traffic, based on the level you set. It is not dynamic: it always duplicates according to the configured level, whether or not conditions currently need it. Using Peplink's own example of a tunnel with three WANs on each side:

  • Off: a packet is carried over 1 tunnel (1x, no duplication).
  • Normal: one duplicate is made, so 2 packets go over 2 tunnels (about 2x the original data).
  • Medium: two duplicates, so 3 packets over 3 tunnels (about 3x).
  • High / Maximum: packets are duplicated across all available tunnels. In the 3-plus-3-WAN example that means up to 9 packets over 9 tunnels, i.e. full duplication (very heavy on data).

In short: the higher the level, the more copies sent and the more data consumed. Plan for this if you're on metered cellular or satellite.

Reducing failover delay with WAN priority

How you place your WANs on the dashboard has a big effect on how quickly the tunnel rides through an outage, especially when mixing a fixed WAN with cellular:

  • Cellular on Priority 2 (standby): the cellular link isn't connected until the primary WAN fails. When it does, the SIM still has to register with the carrier before the tunnel can use it, which adds a noticeable delay to failover.
  • Both WANs on Priority 1 (active): both links stay up. In the SpeedFusion profile, set the cellular link to a lower priority so no user data is pushed over it during normal operation, and the tunnel uses it only for health checks. Because it's already connected, there's no carrier-registration delay when it's needed.

A note on mixed-latency links (cellular + fixed)

If you want minimum packet loss across links with very different latency, use WAN Smoothing on Maximum for full packet duplication across both WANs. This is the right tool for mismatched-latency links because the lower smoothing levels lean on a bonding-style algorithm that you generally do not want to run across connections with very different latency. The trade-off is heavy data usage. If data consumption is a concern instead, use Hot Failover, which conserves data but may drop a few packets during the transition from one WAN to another.

Which should you use?

  • Ultra-low-latency, real-time, must-not-glitch traffic (live video, broadcast, board-room calls, VoIP over shaky links): use WAN Smoothing, and scope it to just those sessions so you're not burning bandwidth on everything.
  • High-latency or very lossy links (LEO satellite, remote cellular), and video streaming or multicast: use FEC, which recovers loss efficiently without costly retransmissions.
  • Both can be layered on a SpeedFusion profile, and you can target them to specific traffic via outbound policy.

Remember these are tunnel features, so they require SpeedFusion or SpeedFusion Connect. For the bigger picture on when to bond at all, see "Bonding vs. Load Balancing," and for setup see "Setting Up SpeedFusion Connect for Bonding and Hot Failover."

Want help tuning a tunnel for VoIP, live video, or satellite? Reach out to sales[at]llamanetworks[dot]com.

Further reading and sources

Was this article helpful?
0 out of 0 found this helpful
Share this article
Have more questions? Submit a request