High Availability Active-Active Deployment Model

High Availability supports the Active-Active deployment model.

AP Load Distribution Across Edge Nodes and Primary Edge Selection by the AP

When the active-active Edge cluster is configured for SD-LAN service, RUCKUS One provides list of access port IPs to each of the APs for the venues which are participating in SD-LAN service. Based on the load distribution algorithm configured in the Active-Active Cluster, APs randomize this list to produce a new randomized list.

To distribute the load of AP tunnels across all nodes in the Active-Active HA mode cluster, two load distribution methods are supported.

  • Random distribution: Each AP randomizes the Edge VTEP IP address list using its serial number as a random seed. Each AP will then attempt to connect to the first Edge IP address in its randomized list, known as the Primary Edge. This approach directs the tunnels from APs to different nodes in the Active-Active HA cluster, effectively distributing the load.
  • Per AP-group distribution: Each AP randomizes the Edge VTEP IP address list using the AP-group identifier. In this method, all APs in the same AP-group will derive the same randomized IP list. As each AP will attempt to connect to the first Edge IP address in its randomized list, all APs in the same AP-group tunnel to same Edge node. As long as a wireless client roams across APs in the same AP-group, there will not be any MAC movement for that wireless client MAC address in the Edge uplink port switch network.

The Edge with first IP address in the list becomes the Primary Edge for the given AP. The Edge to which the AP connects is referred to as the Active Edge. In an ideal scenario, Active and Primary Edge nodes are the same for the AP.

When an AP is connected to a RUCKUS Edge node, it will continuously send periodic keepalive traffic to ensure the Edge device is reachable.

In a High Availability configuration, if an AP is unable to connect to the first Edge IP address in its randomized list, it will attempt to connect to the next IP address in the list and continue this process until a connection is established. If it cannot connect to the last IP address in the list, it will try the first IP address again. The AP will continue this process until it successfully connects to an Edge device.

Fast Tunnel Failover

In addition to the active tunnel, APs can also form a backup tunnel with another Edge in the cluster. After connecting to its Active Edge, the AP connects to next available Edge node in the randomized list to form a backup tunnel. The AP exchanges Keepalive packets with the Backup Edge as well. Client traffic is not forwarded to Backup Edge.

Tunnel Failover Process

If the keepalive message to the Active Edge fails, the AP will attempt to connect to the Backup Edge node. This process is known as AP tunnel failover. When the Primary Edge becomes unavailable, APs seamlessly transition to the backup tunnel. This failover mechanism minimizes service disruption, reducing client connectivity impact and preventing significant traffic loss, thereby ensuring a resilient and highly reliable SD-LAN deployment.

RUCKUS Edge Fallback

The RUCKUS Edge Fallback feature helps balance the load of APs across RUCKUS Edge devices within the Data Center (DC) cluster. The AP will attempt to fall back to the Primary RUCKUS Edge only if it is currently connected to a non-Primary Edge device.

At the scheduled fallback time, the AP tries to set up a tunnel with its Primary Edge; if the tunnel is successfully established, the AP, it immediately switches the active tunnel from the Active Edge to the Primary Edge. In this process, the Backup Edge for the AP may also get reset to the second Edge in the randomized list created by the AP. Refer to RUCKUS Edge Fallback for complete feature details.

Failure and Expected Behavior

Whenever a node fails, all the APs which are tunneling to that node will start tunneling to other available nodes in the cluster.

LAN, Node, and Cluster Link Failure and Expected Behavior - Active-Active HA

Failure Type Failure Behavior Failure Recovery Behavior
Access Port Failure

In this case, APs which are tunneling to this node will start tunneling to other available nodes in the cluster.

After the network or link failure is rectified, the AP may start tunneling to recovered node as a part of the failover or fallback process.
Node Failure

In this case, APs which are tunneling to this node will start tunneling to other available nodes in the cluster.

After the node initializes, it rejoins the cluster.