Failover setup

To run the DCM Dashboard infrastructure in failover a few things have to be taken into account.

Redis External Session Store

To ensure your sessions will remain available when a service goes down a external session store is needed. Every request the session will be retrieved from the external session store, and afterwards it will be stored again in the session store. Because of this we can ensure that, e.g. another runtime can take over the session when another runtime gets unavailable. Sessions from the Runtime and the gateway can be stored in Redis. Services like CDS and Case Engine don't need a session store because they're stateless and thus don't have sessions.

For Runtime Configuration see: 3. Key-value store API and default component

For Gateway : Blueriq Gateway Service

Load balancer

When clustering a lot of information about this can be found on Runtime cluster. In a Failover setup it is very important that every component of the infrastructure is redundant. For connecting these cluster of components load balancers are used to route between the components. Sticky sessions are not needed or desired in a Failover situation, because sessions are stored external, every instance of a component can continue with that stored session. 

For more information : 2. About the load balancer

Keep in mind that configuration, e.g. blueriq.customerdata-client.url, to services should point to the loadbalancer, which is in front of a cluster of services, instead of the direct url of 1 service instance.