HA between UNIStim endpoints or MGC of TDM endpoints and Device Adapter nodes: Intra-cluster redundancy (N+1 clusters)

Last Updated : Dec 14, 2020 |
Prolog information
Metadata
Keywords:
Avaya Breeze® platform supports the Device Adapter solution by using the Avaya Breeze® platform model for intra-cluster redundancy. In this model, endpoints registered with Device Adapter use the cluster IP address of the Avaya Breeze® platform cluster load balancer instead of individual IP addresses of the Avaya Breeze® platform servers.
Intra-cluster redundancy supports a maximum of N+1 configuration in the Avaya Breeze® platform cluster. The maximum number of supported servers in a cluster is 6. This means the maximum useful capacity of a cluster is 5 servers so that the cluster has an N+1 configuration. If a server fails or a network fault isolates the server, the registered endpoints are redistributed to the other servers in the cluster. The cluster uses a leader/follower model for the load balancer and the cluster IP. As a result, if the server that hosts the cluster IP address is unavailable, the cluster IP address is taken over by one of the other servers.
The maximum number of Device Adapter nodes supported in a solution is 50. These nodes are grouped into clusters with no more than six nodes per cluster (High Availability supports up to a 5+1 footprint). Because there can be 1 to 6 nodes per cluster, the maximum number of clusters supported in a solution varies based on the total number of nodes per cluster. However, the maximum number of 50 nodes still places a fixed limit on the number of nodes and users per System Manager. If all clusters are single nodes, there can be no more than 50 clusters defined.
An N+1 cluster configuration allows Device Adapter to be fully functional even if any one cluster encounters an outage or is taken down for maintenance. Call handling and endpoint registrations are performed normally. In an N+1 environment, the load balancer ensures that all nodes have equivalent registration loading. Consequently, different device IP ports can register the device to any of the other five available nodes. However, redundancy is not supported in an N+1 cluster configuration if any one cluster is not functional.
Whereas, an “N+0” cluster configuration provides a smaller footprint for the number of users but does not allow Device Adapter the capability to be fully functional if any one cluster encounters an outage or is taken down for maintenance. For example, if a site has a 5+0 cluster configuration and the services on one node are stopped to perform node maintenance, all the user endpoints on that node are taken out of service, and the endpoints are non-deterministic. Out of the 5 clusters, only 4 clusters are functional. Hence, 20% of the call handling and registration capability is lost until the node under maintenance starts running again.

Fail-over handling by MGC in an intra-cluster redundancy model (N+1 clusters)

MGC follows the following fail-over process in an intra-cluster redundancy model:
  1. During initial registration, the MGC uses the cluster IP address of the Avaya Breeze® platform cluster to register to the Device Adapter node. This cluster IP is configured on the MGC by the system administrator.
  2. If the Device Adapter node does not have the required capacity, the Device Adapter cluster load balancer redirects the MGC to a Device Adapter node that has the required capacity within the cluster. MGC registers the endpoints with the node.
  3. If a connection failure occurs at the node, MGC tries to failover by using the same cluster IP address. The Device Adapter cluster load balancer provides the cluster IP of the cluster that has the required capacity.
  4. MGC registers the endpoints to the new Device Adapter node within the cluster.