Elt

TL;DR

ELT is a three-step data workflow: extract data, load it into a target system, then transform it.
It aims to make data available for analysis quickly and efficiently with minimal upfront preprocessing.
Common in data warehousing and data lake scenarios where transformations like data cleaning, filtering, or aggregation are applied after loading.

Definition

ELT, or Extract, Load, Transform, is a data processing workflow that involves extracting data from one or more sources, loading it into a target system or database, and then transforming it into a format that is suitable for analysis or other purposes. The goal of ELT is to make data available for analysis and decision-making as quickly and efficiently as possible, without the need for extensive data preparation or preprocessing.

Explanation

ELT reverses the traditional order of operations by performing extraction and loading before transformation. Data from various origins is first moved into the target environment, and transformations are applied there to convert the raw data into an analysis-ready form. Typical transformation steps include data cleaning, filtering, and aggregation. This approach emphasizes speed and efficiency in making data accessible for querying and decision-making.

Examples

Data warehousing

Data from a variety of sources, such as transactional databases, flat files, or web applications, is extracted and loaded into a data warehouse. This data is then transformed into a format that is suitable for analysis, for example by applying transformations such as data cleaning, filtering, or aggregation. The transformed data can then be queried using SQL or other tools to generate reports, identify trends, or make data-driven decisions.

Data lakes

Data from a variety of sources is extracted and loaded into a central repository, such as an object store or file system. This data is then transformed into a format that is suitable for analysis, for example by applying transformations such as data cleaning, filtering, or aggregation. The transformed data can then be queried using SQL or other tools to generate reports, identify trends, or make data-driven decisions.

Data warehouse
Data lake
Data cleaning
Filtering
Aggregation
SQL

Elt