Before we go into installing and using Cloudera Manager, you need to understand the architecture of Cloudera Manager and how it functions. Cloudera Manager is composed of the following two parts:
- Cloudera Manager Server
- Cloudera Manager Agent
The following diagram depicts the Cloudera Manager architecture:
The Cloudera Manager Server is the master service that manages the data model of the entire cluster in a database. The data model contains information regarding the services, roles, and configurations assigned for each node in the cluster.
The Cloudera Manager Server is responsible for performing the following functions:
- It communicates with Cloudera Manager Agents that are installed on each node of the cluster and assigns tasks as well as checking the health of each agent by monitoring its periodic heartbeats.
- It provides an administrator web interface for the end user to perform administrator operations.
- It calculates and displays dashboards of the health for the entire cluster.
- It monitors the important parameters such as disk usage, CPU, and RAM for each node in the cluster. It also allows full control on the Hadoop daemons running on the cluster.
- It manages the Kerberos credentials for the services running on the cluster. Kerberos is the tool used to manage the authentication and authorization requirements of the cluster.
- It exposes a set of easy-to-use APIs that helps developers write their own applications to interact with the Cloudera Manager Server.
The Cloudera Manager Agent is installed on each node of the cluster. It is responsible for accepting tasks from the Cloudera Manager Server and performs the starting and stopping of Hadoop daemons on its own node. It is also responsible for gathering all system-level information and statistics and is relayed back to the Cloudera Manager Server.
All communication between the server and its agents is done over HTTP(S). The user also connects to the server using a web browser via HTTP(S). The user uses the web application to perform all the administrator operations.