The Rise of ELT: Cloud Storage and Compute
What you'll learn: How cheap cloud storage and powerful query engines enabled the shift from ETL to ELT workflows.
The Economic and Technical Revolution
Remember from the previous lesson that ETL (Extract, Transform, Load) was the dominant pattern for decades. But two major technological shifts in the 2010s changed everything: cloud storage became remarkably cheap and cloud compute became incredibly powerful.
In the ETL era, storing raw data was expensive—disk space cost dollars per gigabyte. You had to transform data before loading it into your warehouse to minimize storage costs. It was like packing a suitcase carefully before a flight because every pound cost money.
How Cloud Changed the Game
Cloud platforms like AWS, Google Cloud, and Azure introduced:
- Pennies-per-terabyte storage (Amazon S3, Google Cloud Storage): Storing raw data became so cheap that the cost of keeping everything in its original form was negligible
- Massively scalable compute engines (BigQuery, Snowflake, Redshift): These could query petabytes of data in seconds, separating storage from compute power
Why ELT Became Viable
With dirt-cheap storage, you could now:
- Load raw data immediately without expensive pre-processing
- Keep multiple versions and historical snapshots without worry
- Transform data inside the warehouse using SQL, leveraging the warehouse's massive parallel processing power
- Reprocess historical data whenever business logic changed
The transformation step moved from brittle external scripts to flexible SQL queries running on engines designed for exactly this workload.
Key Takeaway: The ELT revolution was enabled by cloud economics—when storage became nearly free and warehouses gained enormous query power, it made more sense to load first and transform later using the warehouse's own scalable compute resources.