This mode is entirely different from Single-Node or Pseudo Distributed modes where only a single node was used.
In Fully Distributed Nodes, multiple nodes (with the minimum recommended 3) work simultaneously to execute Hadoop jobs. All daemons are also services that run on different nodes. The master-slave architecture is also achieved through this, where the master runs on a master node and the slave runs on another node called the slave node. This is how a fully distributed system differs from single-node systems.
To master the fully distributed mode in Hadoop, you can sign up for Intellipaat's Hadoop Online Training.