What is the different between ambari-agent and ambari-server, and why the host where installed ambari have the both(server and agent)

[#1 · Accepted solution]

@yadir Aguilar

Ambari Server is a Java based program which is responsible for managing cluster nodes/ services/components and to monitor them. Ambari Server also provides means to view the cluster details via UI and using API calls. It runs on Jetty and also uses Pythos/Shell scripts for different purposes. Ambari server stores the cluster details and configs inside the Database (by default uses Postgres in Embedded Mode)

On the other hand Ambari Agent is a python script which runs on the Nodes which needs to be part of cluster. So If you want to monitor and manage resources / components on a host then you will need to install Agents on that host. Ambari agents are also responsible to running the scheduled alert scripts to monitor the health of various resources.

The Ambari Server collects data from across your cluster. Each host has a copy of the Ambari Agent, which allows the Ambari Server to control each host. So it is good to have agent installed on ambari server host as well (But not mandatory)

.

[#2]

thanks @Jay Kumar Sen Sharma, i have another question, which are the minimal characteristics if i want to configure a hadoop cluster with ambari on a only one server(hardware characteristics of the server)

[#3]

@yadir Aguilar

For a single node cluster setup with Ambari you will atleast 8 GB RAM so that you can run couple of services (like Yarn/HDFS/MR/Hive...etc) along with the Ambari Server and Agent.
Disk space 10-20 GB should be sufficient.

For Software requirement like Python/Openssl/Operating System details you can get from : https://supportmatrix.hortonworks.com

If you just want to get going with HDP/HDF then HDP Sandbox will be best option: https://hortonworks.com/tutorial/learning-the-ropes-of-the-hortonworks-sandbox/

If you wan to see how a Single Node Cluster installation is done in HDP sandbox (which is a single node cluster setup) then you can take a look at HDP Sandbox

[#4]

@Jay Kumar SenSharma

if i want to configure a data lake for business intelligence, what i have to use, HDP or HDF?