Apache Flume is a highly available, distributed, and configurable tool/service that collects and aggregates large amounts of data such as logs and events from different sources. It is designed to collect data flows (e.g., log data) from various web servers and store them in centralized data storage system like HDFS and HBase.
Flume Architecture
A Flume event is defined as a unit of data flow. A Flume agent is a JVM process that contains the components required for completing a task. Among them, Source, Channel, and Sink are the core ones.
Source
It consumes an event passed to it by an external source (e.g., web servers or other sources) and save it to one or more channels.
Channel
Located between a source and a sink, a channel is used to cache incoming events. After the sink successfully sends the events to the channel at the next hop or the final destination, the events are removed from the channel.
Sink
A sink is responsible for transferring the events to the next hop or final destination and removing them from the channel upon transfer completion.
Instructions
Preparations
Create an EMR cluster. When creating the EMR cluster, you need to select the Flume component on the software configuration page.
Install Flume in the /usr/local/service/flume path on the core and task nodes (CVM instances) of the EMR cluster. The installation path for master nodes is /usr/local/service/apps/.
Configuring Flume
Go to the /usr/local/service/flume folder and create a file named example.conf.
