The folks at Data Flair have a tutorial on how Ambari is architected:
Ambari Architecture is of master/slave type architecture. So, to perform certain actions and report back the state of every action, the master node instructs the slave nodes. Although, for keeping track of the state of the infrastructure, the master node is responsible. But for this process, a database server is used by the master node, that can be further configured during setup time.
Now, we can see the high-level architecture of Ambari by below diagram which also shows how Ambari works:
Ambari is one of the easiest ways I’ve seen to spin up and manage a Hadoop cluster.