Oracle added 'Analytic Views' feature in its database in 2017, starting with version 12.2 of database. It evolved from OLAP engine inside Oracle database, which used to be 'EXPRESS' multidimensional engine in 1980s and 1990s.
To me this is just lingo (or should I say just semantics). Data Dictionaries along with Conceptual or Logical Data Models basically serve the same purpose. Tell me if I am wrong.
I really enjoyed the end: without a match between technology and stakeholders, engineers cannot do much! The semantic layer tools can only alleviate this burden... hopefully!
An important point that I think you've missed, my friend:
The Semantic Layer was indeed vendor lock-in BS as you described.
However, with the emergence of Natural Language BI as the holy grail, it appears there's no trustworthy way of launching it without some Semantic Layer that feeds the agent.
Good explanation overall. I think the easiest way to understand a semantic layer is:
👉 It acts like a “business translation layer” between raw technical data and business users.
Instead of forcing teams to understand complex tables, joins, and SQL logic, the semantic layer gives everyone a common business language for metrics like revenue, customers, churn, or active users.
Without a semantic layer:
finance, sales, and marketing may all calculate the same KPI differently
dashboards start showing conflicting numbers
trust in data drops quickly
With a semantic layer:
metrics become standardized
business logic becomes reusable
self-service analytics becomes easier
AI and BI tools get more reliable context
One Reddit comment explained it well:
“The semantic layer is the translation of raw data into usable data.”
I also think semantic layers are becoming even more important now because AI systems and copilots need consistent business definitions to avoid hallucinations and incorrect analytics.
Great explanation of the semantic layer and its growing importance in modern data architectures. I especially liked how the article highlights the need for a consistent business definition layer, which is critical as organizations deal with multiple data sources and AI-driven analytics. A semantic layer essentially bridges the gap between raw data and business users by translating complex data into meaningful insights, enabling better decision-making and reducing confusion across teams .
With the rise of AI and self-service analytics, having a well-defined semantic layer is no longer optional it’s becoming a foundational component for data governance and scalability. Looking forward to seeing more real-world implementations and best practices in this space!
I really should read the rest (skimmed), but you lost me at this point in terms of an argument. These aren't definitions that "differ significantly" IMO. One is more broad/general (DuckDB) than the other (Databricks).
Interestingly, if you are an astute observer, you will notice that DuckDB and Databricks actually differ significantly in their definitions of a Semantic Layer.
Databricks says it’s AFTER Data Lakes, Data Marts, but BEFORE the BI tools.
DuckDB says it’s simply something AFTER the database and BEFORE the business user.
popcorn as a poll option is my top highlight
Oracle added 'Analytic Views' feature in its database in 2017, starting with version 12.2 of database. It evolved from OLAP engine inside Oracle database, which used to be 'EXPRESS' multidimensional engine in 1980s and 1990s.
As I have worked with Snowflake Semantic Views and Oracle 'Analytic Views' (AV), I have found that Oracle AV syntax is much easier and it has more features than Snowflake Semantic Views. https://docs.oracle.com/en/database/oracle/oracle-database/23/dwhsg/overview-analytic-views.html There are some blogs and YouTube videos about this excellent feature also.
To me this is just lingo (or should I say just semantics). Data Dictionaries along with Conceptual or Logical Data Models basically serve the same purpose. Tell me if I am wrong.
I really enjoyed the end: without a match between technology and stakeholders, engineers cannot do much! The semantic layer tools can only alleviate this burden... hopefully!
hilarious
An important point that I think you've missed, my friend:
The Semantic Layer was indeed vendor lock-in BS as you described.
However, with the emergence of Natural Language BI as the holy grail, it appears there's no trustworthy way of launching it without some Semantic Layer that feeds the agent.
No wonder Snowflake tries to standardize this mess into a single semantic syntax - https://www.snowflake.com/en/blog/open-semantic-interchange-ai-standard/
Good explanation overall. I think the easiest way to understand a semantic layer is:
👉 It acts like a “business translation layer” between raw technical data and business users.
Instead of forcing teams to understand complex tables, joins, and SQL logic, the semantic layer gives everyone a common business language for metrics like revenue, customers, churn, or active users.
Without a semantic layer:
finance, sales, and marketing may all calculate the same KPI differently
dashboards start showing conflicting numbers
trust in data drops quickly
With a semantic layer:
metrics become standardized
business logic becomes reusable
self-service analytics becomes easier
AI and BI tools get more reliable context
One Reddit comment explained it well:
“The semantic layer is the translation of raw data into usable data.”
I also think semantic layers are becoming even more important now because AI systems and copilots need consistent business definitions to avoid hallucinations and incorrect analytics.
https://www.solix.com
Great explanation of the semantic layer and its growing importance in modern data architectures. I especially liked how the article highlights the need for a consistent business definition layer, which is critical as organizations deal with multiple data sources and AI-driven analytics. A semantic layer essentially bridges the gap between raw data and business users by translating complex data into meaningful insights, enabling better decision-making and reducing confusion across teams .
With the rise of AI and self-service analytics, having a well-defined semantic layer is no longer optional it’s becoming a foundational component for data governance and scalability. Looking forward to seeing more real-world implementations and best practices in this space!
https://atechreview.com/
I really should read the rest (skimmed), but you lost me at this point in terms of an argument. These aren't definitions that "differ significantly" IMO. One is more broad/general (DuckDB) than the other (Databricks).
Interestingly, if you are an astute observer, you will notice that DuckDB and Databricks actually differ significantly in their definitions of a Semantic Layer.
Databricks says it’s AFTER Data Lakes, Data Marts, but BEFORE the BI tools.
DuckDB says it’s simply something AFTER the database and BEFORE the business user.