For a third-party load balancer to route traffic correctly between two or more Context Store clusters, the following mandatory steps must be configured on the load balancer:
Ensure that the traffic is routed on the rid parameter.
Each request for a particular context must be routed to the same Context Store cluster. To achieve this, Context Store uses an optional field, rid parameter, in each REST request. For example, all requests with rid parameter=1 must be directed to the nondefault cluster and any other requests must be sent to the default cluster. Requests for the same context must go to the same Context Store cluster as the time lag for replication between the two Context Store clusters is not instantaneous. You cannot route requests using round robin, least busy server, or any other mechanism other than routing on the rid parameter. You must ensure that each request for a particular context has the same rid parameter.
If the load balancer receives responses with any of the following codes: 500, 502, 503, and 504, the load balancer must mark the Context Store cluster as down and redirect all subsequent requests to the other Context Store cluster. The 500 code signifies that only one of the Avaya Breeze® platform nodes in the Context Store cluster is operational and it cannot process requests.
The load balancer must be able to forward the Context Store https requests. If the load balancer requires authentication, load the certificates in the load balancer. For more information about certificates, see Certificate-based authentication.
