Traffic between the clusters 

Traffic between the edge cluster and the hub cluster is managed with permits.

Traffic between the edge cluster and the hub cluster is managed with permits.

Communication traffic priorities

All messages that will be replicated are stored in the replication log. When the replication log is queried, messages are processed according to the time and priority of the replication channel. Replication channel priorities determine whether the message in the replication log travels to the hub by the fast track or the slow track.
Fast track

Replication channels that are configured with priority 1 always acquire permits, if available, and use the fast track to route messages. These messages are stored in the replication log on the edge and are then immediately sent to the hub. Fast track messages will block waiting for an available permit. When permits become available, the permits are served in the order they were requested (first come, first served). A bounded in-memory queue is managed to prevent the waiting queue from growing indefinitely. Messages that do not fit into the bounded queue are replicated with the slow track.

For example, to create a replication channel with priority 1 to route messages on the fast track:
$ dse advrep -q --host 192.168.3.10 edge channel create --keyspace "demo" --table "mytable" --priority 1
Slow track

Replication channels that are configured with priorities other than 1 acquire permits if available, and if no fast track messages are present. Channels with priorities other than 1 use the slow track to route messages. Slow track messages wait for a configurable amount of time until a permit becomes available. If the configured time elapses without acquiring a permit, the messages are discarded and waits in the replication log on the edge cluster to await processing to hub.

For example: to create a replication channel with priority 2 to route messages on the slow track:
$ dse advrep -q --host 192.168.3.20 edge channel create --keyspace "demo2" --table "mytable2" --priority 2

Permits  

Traffic between the edge cluster and the hub cluster is managed with permits. Permits limit the number of concurrent operations that are submitted to the Cassandra driver. A permit must be acquired before a message can be sent to the hub. When a permit cannot be acquired, the message is discarded and waits in the replication log until it is processed when a permit becomes available.

To manage bandwidth, you can use these parameters to tune the permit configuration:
replication-max-permits
The maximum number of permits that are available for replication.
replication-min-permits
The minimum number of permits that are available for replication.
slow-track-min-bandwidth-percentage
The minimum guaranteed percent of permits that are available for sending messages by using the slow track.
replication-permit-timeout-ms
If permits are not available, the maximum time the replication channel waits to acquire a permit. Applies only to messages sent by the slow track.

Configure automatic failover for hub clusters with multiple datacenters

DSE Advanced Replication uses the Java driver load balancing policy to communicate with the hub cluster. You can explicitly define the local datacenter for the datacenter-aware round robin policy (DCAwareRoundRobinPolicy) that is used by the Java driver.

You can enable or disable failover from a local datacenter to a remote datacenter. When multiple datacenter failover is configured and a local datacenter fails, data replication from the edge to the hub continues using the remote datacenter. Tune the configuration with these parameters:
driver-local-dc
For hub clusters with multiple datacenters, you can explicitly define the name of the datacenter that you consider local. Typically, this is the datacenter that is closest to the edge cluster. This value is used only for clusters with multiple data enters.
driver-used-hosts-per-remote-dc
To use automatic failover for hub clusters with multiple datacenters, you must define the number of hosts per remote datacenter that the datacenter aware round robin policy (DCAwareRoundRobinPolicy) considers available for use.
driver-allow-remote-dcs-for-local-cl
Set to true to enable automatic failover for hub clusters with multiple datacenters. The value of the driver-consistency-level parameter must be LOCAL_ONE or LOCAL_QUORUM.