uses the round-robin method to distribute the messages across the instances.
EDS
avoids data duplication within the load balanced group of instances by delivering each message to one instance. Deploying a target service on multiple nodes ensures that a large volume of data from multiple sources does not overwhelm a single target service instance.
The following image shows how
EDS
balances the load across HDFS target services that you deployed on three nodes:
EDS
balances the load as follows:
A source service reads data from the data source and publishes it as three messages over a topic called logs.
Three instances of a target service receive the messages.
The target service is deployed on three nodes for purposes of load balancing.
EDS
balances the load across the three instances of the target service in round-robin fashion.
Although
EDS
uses the round-robin method, some instances might receive more messages than other instances.
If one or more instances become inactive,
EDS
removes the inactive instances from load balancing decisions and redistributes the load across the remaining instances.