In 3 Tier Layer with ELT (NOT ETL), the more I research, the more I get confused.
Can you explain layer by layer and their differences?
​
Landing: Landed from source systems into tier 1?
Staging: After landing, pre filter rules before moving to tier 2?
Single Image: Staging + Bkey + Bmap fields?
Mod (Tier 2): Does it stand for Modular or Module? What is this layer exactly?
Single Event Repository: Enterprise event storage? What is the purpose of this layer if source system and landing layers are already there? Is this Tier 1 or 2?
Semantic (Tier 3): Business layer before app layer so people can tweak the logic before shown in the app layer?
In: Engineering
You might be referring to a specific implementation that is not common to all DW. That said, here are some general descriptions.
Layer 1 is raw, unmodified, unfiltered data. It is everything you receive in the same format as you receive it, or as close as possible. It allows you to build, and rebuild if there was some error, the later layers.
Layer 2 will be for filtered, modified, restructured, and calculated data. The mod here is likely ‘modified’ as opposed to ‘raw’. This is where you apply the T of ELT, the structure is optimised for easier and faster transformation even if it makes it harder to analyse by business.
Layer 3 is for the consumers. You take the transformed data from Layer 2 and restructure it towards analytical needs. Everything here is designed for fast and easy analytical queries for business problems.
Latest Answers