TEZ-431. Implement fault tolerance, retries and event flow for dealing with failed inputs (bikas)
30 files changed
tree: 36136da2451744bb2c84e0125f6ed4e2abe9e47d
  1. .gitignore
  2. INSTALL.txt
  3. README.md
  4. docs/
  5. example_jobs/
  6. ljr_helper/
  7. pom.xml
  8. tez-api/
  9. tez-common/
  10. tez-dag/
  11. tez-dist/
  12. tez-mapreduce-examples/
  13. tez-mapreduce-tests/
  14. tez-mapreduce/
  15. tez-runtime-internals/
  16. tez-runtime-library/
  17. tez-yarn-client/
README.md

tez

tez is a generic data-processing pipeline engine envisioned as a low-level engine for higher abstractions such as Apache Hadoop Map-Reduce, Apache Pig, Apache Hive etc.

At it's heart, tez is very simple and has just two components:

  • The data-processing pipeline engine where-in one can plug-in input, processing and output implementations to perform arbitrary data-processing. Every ‘task’ in tez has the following:
  • Input to consume key/value pairs from.
  • Processor to process them.
  • Output to collect the processed key/value pairs.
  • A master for the data-processing application, where-by one can put together arbitrary data-processing ‘tasks’ described above into a task-DAG to process data as desired. The generic master is implemented as a Apache Hadoop YARN ApplicationMaster.