Backgroud

This project is aiming to build a whole cloud based DevOps ETL process. Include below Parts:

AWS

1. Cloud Infrastructure
   • Jenkins on ECS
   • Airflow on EKS
2. Airflow framework(wrapper)
3. Jenkins Devops Pipeline
4. Glue ETL Common Solution
5. Multi-account architecture

Power BI

1. Front end development & design
2. Backend development & design
3. DB development & design

Azure

1. User/Role Management Architecture
2. Network/Security Architecture
3. DevOps Architecture
   • Infrastructure Level DevOps
   • Project Level DevOps
4. Project Architecture
   • ETL framework/solution
   • Data Visualization(PowerBI)

Project Name

Cloud base ETL DevOps process of Community = CEDC

Project Directory

• Airflow Workspace
• Glue Workspace
• Jenkins Workspace
  • Jenkins Infrastructure
  • Jenkins Pipelines
• Sample Code

Project Wiki

Project Wiki

• Home
• Q&A

Project Sprint

Sprint

Architecture

basic logicflow

Cloud Infrastructure

Account distribution

• DevOps Account: this is a DevOps account mainly include Jenkins and Airflow
• Data Account: this is a data lake account mainly include S3
• Serverless Account: this is a ETL account mainly include Glue, Lambda etc
• IDP Account: this is a Identity account which can assume A/B/C accounts by User role or Admin Role

jenkins Infrastructure

Note: in the first draft, we can centralized deploy all services into one account for demo purpose.

Airflow framework

Features

• Parameter driven framework
• Check Dependence
• Kickoff
• Monitor
• Job Retry
• Notify
• Metadata backend

Jenkins DevOps Pipeline

Features

• Deploy airflow dags and glue job in project
• Onboarding/Off Boarding
• Data validation
• Convert SQL to Glue Pyspark

Glue ETL jobs

Account prerequisite

Standard aws serverless account with below items:

• Glue
• Lambda
• S3
• Cloudwatch Events
• Cloudwatch logs
• Secrets manager
• wip ...

Glue

Glue job naming standard:

• __prelanding
• __landing
• __landing_merge
• __refinement
• __publish

IAM Roles Management

1. Serverless Account: Glue Job Execution role -> DEVOPS_GLUE_CEDC_EXECUTION (cross account role to ensure Airflow can trigger glue jobs on Account C)
2. DevOps Account: DEVOPS_GLUE_CEDC_READ/DEVOPS_GLUE_CEDC_ADMIN (Readonly or Admin)
3. IDP Account: CICD Role: DEVOPS_CICD_CEDC (which will assume admin access for all accounts for now.)
4. Data Account: DEVOPS_S3_CEDC_READ/DEVOPS_S3_CEDC_ADMIN

OpenAI