Intro to Big Data AppHub: Demo of Kafka to HDFS and HDFS to HDFS templates

Abstract:
To make critical business decisions in real time, many businesses today rely on a variety of data, which arrives in large volumes. Variety and volume together make big data applications complex operations. Big data applications require businesses to combine transactional data with structured, semi-structured, and unstructured data for deep and holistic insights.

And, time is of the essence: to derive the most valuable insights and drive key decisions, large amounts of data have to be continuously ingested into Hadoop data lakes as well as other destinations. As a result, data ingestion poses the first challenge for businesses, which must be overcome before embarking on data analysis.

With its various Application Templates for ingestion, DataTorrent allows users to: Ingest vast amounts of data with enterprise-grade operability and performance guarantees provided by its underlying Apache Apex framework. Those guarantees include fault tolerance, linear scalability, high throughput, low latency, and end-to-end exactly-once processing. Quickly launch template applications to ingest raw data, while also providing an easy and iterative way to add business logic and such processing logic as parse, dedupe, filter, transform, enrich, and more to ingestion pipelines. Visualize various metrics on throughput, latency and app data in real-time throughout execution.

Template descriptions:
Kafka to HDFS: The Kafka to HDFS Application Template continuously reads messages from configured Apache Kafka topic(s) and writes each message as a line in Hadoop HDFS file(s).

HDFS Line Copy: The HDFS to HDFS Line Copy Application Template continuously ingests files as lines from between Hadoop clusters, retaining one-to-one file traceability

Presenters:
Ashwin Putta, Product Manager at DataTorrent

Sanjay Pujare, Engineer at DataTorrent

Dr. Munagala V. Ramanath ("Ram"), Committer for Apache Apex and Software Engineer at DataTorrent