Skip to content

Latest commit

 

History

History
65 lines (62 loc) · 2.03 KB

README.md

File metadata and controls

65 lines (62 loc) · 2.03 KB
Raw

AcidOnSpark-ETL

Watch about it here:

youtube

Read about it here:

medium

Technologies

The main technologies are:

  • Python3
  • Docker
  • Spark
  • Airflow
  • MinIo
  • DeltaTable
  • Hive
  • Mariadb
  • Presto
  • Superset

How is it work?

I this section each part of this ETL pipeline will be illustrated:

Spark

Spark is used to read and write data in distributed and scalable manner.

make spark

will run spark master and one instance of worker

make scale-spark

will scale spark worker.

Airflow

One of the best workflow managemnet for spark jobs.

make airflow

MinIo

An opensource, distributed and performant object storage for datalake files and hive tables.

make minio

DeltaTable (Deltalake)

An opensource columnar parquet files formats with snappy compression. Delta supports update and delete, which is very nice. All necessary jar files for supporting delta and s3 objects are added to hive and spark docker images.

Hive and Mariadb

In order to create tables to run Spark SQL on delta tables, spark needs hive metastore and hive needs mariadb as metastoreDb. Mariadb is also used for data warehouse for to run query faster and create dashboards.

make hive

It will create hive and mariadb instances.

Presto

In order to have acces to delta tables without spark, presto is going to be employed as distributed query engine. It works with superset and hive tables. Presto is opensource, scalable and it can connect to any databases.

make presto-cluster

By this command will create a presto coordinator and worker, the worker can scale horizontally. In order to query delta tables using presto:

make presto-cli

In presto-cli just like spark sql, any query can be run.

Superset

Superset is opensource, supports any databases with many dashbord styles also famous in tech in order to create dashboards or to get hands on databases.

make superset