Airflow at UniCredit: Our journey from mainframe scheduling to modern data processing

Опубликовано: 13 Октябрь 2023
на канале: Apache Airflow
371
6

Presented by Jan Pawlowski & Jędrzej Matuszak at Airflow Summit 2023.

Representing the Murex Reporting team at UniCredit we would like to present our journey with Airflow, and how over the past two years it enabled us to automate and simplify our batch workflows. Comparing to our previous rigid mainframe scheduling approach, we have created a robust and scalable framework complete with a CI/CD process, bringing our time to market of scheduling changes down from 3 days to 1. Basing our solution on DAG networks joined by ResumeDagRunOperators and an array of custom-built plugins (such as static time predecessors) we were able to replicate the scheduling of our overnight ETL processes (consisting of approx.