MultiWAN design and configuration guide

This guide explains how MultiWAN in SC//AcuVigil™ handles automatic failover, load balancing, and unstable WAN links. It's for IT staff who plan and configure sites with multiple WAN connections.

MultiWAN monitors each WAN interface and routes traffic through the best available connection, based on its health, its configuration, and its priority. Load balancing groups and interface weights define how traffic is shared across links and how failover happens when links go offline. Lower-numbered groups have higher routing priority, and the weights within each group set how much traffic each interface carries.

The Sensitivity setting controls how closely MultiWAN monitors each WAN interface. It can be set to High, Medium, or Low. Each level sets how often checks run and how many failures or successes in a row mark a link offline or online. Sensitivity affects how fast failover happens, how well MultiWAN ignores brief problems, and how often failback delay and hysteresis come into play.

To limit unnecessary switching caused by flapping links (connections that go offline and online repeatedly), MultiWAN uses two mechanisms:

• Failback delay: When a failed WAN interface comes back online, MultiWAN can wait before making it the preferred path again. This avoids routing changes caused by links that are only briefly healthy.
• Hysteresis: If a WAN interface flaps several times within a short time window, MultiWAN pauses failover decisions for that interface. This gives the connection time to stabilize.

Features

Routing and load balancing

• Automatic failover: When a WAN connection goes down, MultiWAN switches to the next available interface.
• Outbound load balancing: MultiWAN distributes outbound, non-VPN traffic across multiple WAN interfaces. It balances by connection; it doesn't bond channels.
• Load balancing groups: WAN interfaces are grouped to set routing and failover priority. A lower group number means a higher priority.
• Interface weighting: Weights set the share of traffic each interface carries within its load balancing group.
• Policy and rule-based routing: Policies group interfaces and set fallback behavior. Rules decide which traffic uses a policy, based on source and destination IP addresses, ports, or protocols.
• VPN precedence: Interface priority decides which interface builds VPN tunnels. VPN traffic itself isn't load balanced.

Health monitoring and stability

• Active health checks: MultiWAN tests each WAN interface with UDP tests (the default) or ping (ICMP) tests against tracking IP addresses.
• Evaluation strategies: Health check results can be evaluated with different pass strategies, such as reliability, all, any, or majority.
• Sensitivity presets: High, Medium, and Low presets set the check interval and the number of successful or failed checks in a row that mark a link online or offline.
• Failback delay: A waiting period before MultiWAN reactivates an interface that has just recovered.
• Hysteresis (flap protection): A pause in failover decisions for an interface that goes offline and online repeatedly within a time window.
• Hotplug event handling: MultiWAN can either act on or ignore physical plug and unplug events on the interfaces.

The Settings sub-view of the site's MultiWAN tab shows Use Ping Tests, Sensitivity, Failback Delay, Track IP, the hysteresis fields, and the Weight and Load Balance Group for each interface. For the steps, see Configuring the MultiWAN application.

Sensitivity in MultiWAN

Sensitivity sets how quickly MultiWAN detects that a WAN interface has gone offline or come back online, so it sets how fast failover and failback happen. Choose a level that balances responsiveness against stability, based on how reliable your WAN connections are and how much brief disruption you can tolerate.

What sensitivity controls

Sensitivity sets three values:

• Interval: How often MultiWAN checks the health of a WAN interface.
• Down count: How many failed checks in a row mark the interface offline.
• Up count: How many successful checks in a row mark the interface online again.

Together, these values set the effective failover and recovery time. A shorter interval and lower counts react faster to changes. A longer interval and higher counts tolerate more brief or random packet loss.

Sensitivity levels

How it works

MultiWAN checks every WAN interface at the selected interval. When an interface fails the down count of checks in a row, it's marked offline. When it passes the up count of checks in a row, it's marked online again.

• Up time = up count × interval. For example, 3 × 180 = 540 seconds for Low sensitivity.
• Down time = down count × interval. For example, 2 × 120 = 240 seconds for Medium sensitivity.

These times control how long MultiWAN waits before it fails over to another WAN or restores a link that failed.

Choosing the right sensitivity

• High suits very stable environments where uptime is critical and you need a quick reaction. It's more likely to fail over because of brief packet loss or jitter.
• Medium is a practical balance for most networks. It reacts to real problems while filtering out noise.
• Low suits networks where some packet loss is normal, or where you prefer stability over quick recovery. It avoids failovers caused by brief or minor disruptions.

Relation to failover behavior

Sensitivity also affects failback delay and hysteresis. For example, High sensitivity can trigger hysteresis more often when a link flaps, while Low sensitivity tolerates more instability before reacting.

Interface weights and load balancing groups

MultiWAN uses load balancing groups and interface weights together to manage how internet traffic is routed.

Load balancing group priority

Each WAN interface belongs to a load balancing group, which sets its priority. MultiWAN always prefers interfaces in the group with the lowest number: a lower group number means a higher priority.

For example, interfaces in group 1 have priority over interfaces in group 2. While at least one interface in the highest-priority group is online, all traffic is routed within that group. If every interface in that group goes offline, MultiWAN automatically fails over to the next group in priority order.

Interface weight within a group

Within each load balancing group, interfaces can have different weights, which control how traffic is shared among them. An interface with a higher weight carries a larger share of the traffic.

The weights within a group must add up to 100, and the minimum weight is 20. The Weight tooltip on the Settings sub-view says: "Weights within a Load Balance Group must add up to 100. Minimum valid weight is 20."

For example, if two interfaces in the same group have weights of 33 and 67, the second carries about twice as much traffic as the first. If all interfaces in a group have the same weight, traffic is evenly distributed among them, for example 50 and 50.

If a group has only one interface, it has a weight of 100.

How it all comes together

MultiWAN routes traffic in two steps:

1. It finds the highest-priority (lowest-numbered) group that has at least one working interface.
2. It distributes traffic among the interfaces in that group according to their weights.

If every interface in the active group goes down, MultiWAN fails over to the next group and distributes traffic within that group based on its weights.

Hysteresis

When a WAN link goes down, MultiWAN switches traffic to another link. Sometimes, though, a WAN interface doesn't go down and stay down: it goes offline and online repeatedly within a short time. This is called flapping, and it can cause unnecessary failovers and network instability.

Hysteresis prevents this by adding a cooling-off period when MultiWAN detects rapid flapping. Instead of reacting to every down event, MultiWAN pauses failovers for the unstable interface and gives it time to settle.

What triggers hysteresis

MultiWAN monitors each WAN interface with health checks. When a link fails several checks in a row, it's considered offline. When it passes several in a row, it's online again.

1. Tracking transitions: MultiWAN records each time an interface goes from online to offline.
2. Sliding time window: MultiWAN counts how many of these transitions happened within a recent time window.
3. Trigger condition: If the number of transitions reaches the configured threshold, hysteresis is triggered. MultiWAN pauses failover decisions for that interface for a set period.

Hysteresis settings

These fields are on the Settings sub-view of the site's MultiWAN tab.

What happens when hysteresis is triggered

• MultiWAN skips failover decisions for that interface, even if the interface comes back online.
• The pause gives MultiWAN time to see whether the interface stabilizes before using it again.

When hysteresis isn't applied

Hysteresis isn't triggered when:

• Only one WAN interface is online. No failover is possible, so hysteresis is skipped.
• The interface hasn't gone from online to offline (no new flap).
• Hysteresis is Disabled, or Failure Count is 0.

Manual recovery

Hysteresis clears automatically after the pause period. Scale Computing Support can also resume an interface manually on the ESD.

Why hysteresis matters

Without hysteresis, a flapping WAN interface can cause constant failover and failback, interrupted sessions, and unstable routing. Hysteresis adds a pause before MultiWAN acts again, which leads to fewer false failovers and smoother network transitions.

Failback delay

Failover and failback happen automatically as WAN interfaces go offline and come back online. Bringing an interface back into use (failback) too quickly can cause problems, especially if the interface is flapping.

Failback delay keeps routing stable by adding a short wait before MultiWAN reactivates an interface that has just come back online.

What happens when an interface comes back online

When an interface goes from offline to online, MultiWAN decides whether to use it again immediately or to wait for the failback delay first.

When the interface is used again immediately

MultiWAN reactivates the interface right away when:

1. No failback delay is configured. There's nothing to wait for.
2. The interface has no recorded offline time. This usually means the interface is coming online for the first time since startup, so MultiWAN treats it as a fresh start.
3. The delay has already passed. If the interface was offline for longer than the configured delay, it's ready to use.
4. A higher-priority interface, or an interface in the same group, is already online. Using this interface won't disrupt traffic, so there's no need to wait.
5. No other interfaces are online. The interface must be used right away to restore connectivity.

When failback delay applies

Failback delay applies when another interface is already online and that interface is in a lower-priority group (a higher group number).

The delay doesn't apply between interfaces in the same load balance group.

Failback delay settings

Why failback delay matters

Without failback delay, an interface that has just recovered can take over traffic immediately, even if it fails again a few seconds later. That causes frequent route changes, packet loss, and reconnections. By waiting, MultiWAN makes sure interfaces are stable before it uses them, keeps routing consistent, and reduces interruptions to network sessions.