Historical Context: Why ETL Came First
You'll learn: Why ETL (Extract, Transform, Load) became the standard approach before ELT emerged as an alternative.
The Storage-and-Compute Economics of the Past
In the 1990s and early 2000s, storage was extremely expensive. Data warehouses like Oracle, Teradata, and IBM DB2 charged premium prices per gigabyte. Companies couldn't afford to store messy, raw data in these systems—every byte counted.
At the same time, warehouse compute power was limited and costly. Running complex transformations inside the warehouse consumed precious CPU cycles and slowed down business-critical queries. The solution? Do the heavy lifting before loading data into the warehouse.
Why Transform Before Loading Made Sense
Think of it like moving into an expensive apartment: you wouldn't bring all your boxes, junk, and unsorted belongings into a tiny, premium space. Instead, you'd clean, organize, and downsize in your garage first, then only move in what you truly need.
ETL worked the same way:
- Extract data from source systems
- Transform it on cheaper servers or ETL tools (like Informatica or DataStage)
- Load only clean, aggregated, ready-to-query data into the warehouse
This minimized warehouse storage costs and kept compute resources free for end-user queries rather than transformation jobs.
The Economic Logic
ETL dominated because the cost structure favored external transformation. Companies invested in dedicated ETL servers and middleware because it was far cheaper than expanding warehouse capacity or slowing down analytical queries with transformation workloads.
Key Takeaway: ETL became the standard because expensive storage and limited warehouse compute made it economically necessary to transform data before loading it into the data warehouse.