METL: a modern ETL pipeline with a dynamic mapping matrix

Modern ETL streaming pipelines extract data from various sources and forward it to data warehouses (DW) and systems that leverage machine learning (ML). However, the increasing number of systems that are connected to such pipelines requires new solutions for data integration. The canonical, or common, data model (CDM) offers such an integration. It is particular useful for pipelines with microservices. (Villaca et al 2020, Oliveira et al 2019) However, a mapping to a CDM is complex. (Lemcke et al 2012) The mapping requires a parameter matrix. Since the schemata that describe the extracted data can expand with new versions, so does the mapping matrix. We present a new CDM mapping for an ETL pipeline at EOS that uses log-based Change Data Capture (CDC) to extract schematized data from more than 80 microservices. We show that the mapping matrix can quickly grow to up to 1.000.000.000 elements. The size makes updates of the matrix and the computation of the mappings extremely cumbersome. In the paper, we present a new solution for this complexity problem. More precisely, we present an app called Message ETL (METL) that is driven by a new dynamic mapping matrix (DMM). The DMM is based on permutation matrices that are obtained by block-partitioning and pattern generalization. We show that the DMM can be used for automated updates, for parallel mapping in near real-time and for efficient compacting. For the solution, we draw on research into matrix partitioning (Quinn 2004) and dynamic networks (Haase et al 2021). The app with the DMM has been implemented into the new ETL pipeline at EOS. It uses Kafka-streams and Debezium for CDC. EOS is part of the Otto-Group, the second largest e-commerce provider in Europe.