内容简介:WORK-IN-PROGRESSStreaming data changes to a Data Lake with Debezium and Delta Lake pipelineThis is an example end-to-end project that demonstrates the Debezium-Delta Lake combo pipeline
WORK-IN-PROGRESS
delta-architecture
Streaming data changes to a Data Lake with Debezium and Delta Lake pipeline https://medium.com/@yinondn/streaming-data-changes-to-a-data-lake-with-debezium-and-delta-lake-pipeline-299821053dc3
This is an example end-to-end project that demonstrates the Debezium-Delta Lake combo pipeline
See medium post for more details
High Level Strategy Overview
- Debezium reads database logs, produces json messages that describe the changes and streams them to Kafka
- Kafka streams the messages and stores them in a S3 folder. We call it Bronze table as it stores raw messages
- Using Spark with Delta Lake we transform the messages to INSERT, UPDATE and DELETE operations, and run them on the target data lake table. This is the table that holds the latest state of all source databases. We call it Silver table
- Next we can perform further aggregations on the Silver table for analytics. We call it Gold table
Components
- compose: Docker-Compose configuration that deploys containers with Debezium stack (Kafka, Zookeepr and Kafka-Connect), reads changes from the source databases and streams them to S3
- voter-processing: Notebook with PySpark code that transforms Debezium messages to INSERT, UPDATE and DELETE operations
- fake_it: For an end-to-end example, a simulator of a voters book application's database with live input
Instructions
Start up docker compose
- export DEBEZIUM_VERSION=1.0
- cd compose
- docker-compose up -d
Config Debezium connector
curl -i -X POST -H "Accept:application/json" -H "Content-Type:application/json" http://localhost:8084/connectors/ -d @debezium/config.json
Run spark notebook
Import the notebook file in \voter-processing\voter-processing.html to a Databricks Community account and follow the instructions inside the notebook
https://community.cloud.databricks.com/
TODO - To complete the end-to-end example flow
- Change the voter-processing from notebook to PySpark application
- Add the PySpark application to the Docker-Compose
- Change the configurations so that Kafka writes to local file system instead of S3
- Change the Spark application so that it read Kafka's output instead of generating it's own mock data
What's Next?
Make it a configurable generic tool that can be assembled on top of any supported database
以上就是本文的全部内容,希望对大家的学习有所帮助,也希望大家多多支持 码农网
猜你喜欢:本站部分资源来源于网络,本站转载出于传递更多信息之目的,版权归原作者或者来源机构所有,如转载稿涉及版权问题,请联系我们。
编程之美:微软技术面试心得
《编程之美》小组 / 电子工业出版社 / 2018-9 / 79
《编程之美:微软技术面试心得》收集了约60道算法和程序设计的题目,这些题目大部分在微软的笔试、面试中出现过,有的曾被微软员工热烈地讨论过。作者试图从书中各种有趣的问题出发,引导读者发现问题、分析问题、解决问题,寻找更优的解法。《编程之美:微软技术面试心得》内容分为以下几个部分。 游戏之乐:从游戏和其他有趣问题出发,化繁为简,分析总结。 数字之魅:编程的过程实际上就是和数字及字符打交道的......一起来看看 《编程之美:微软技术面试心得》 这本书的介绍吧!
XML 在线格式化
在线 XML 格式化压缩工具
RGB HSV 转换
RGB HSV 互转工具