Module 1 · Lecture 1.4
Inside the Knowledge Graph & Database Independence
Two different databases, one identical VQL query: why the Knowledge Graph is independent of the database underneath.
This lecture looks inside the Knowledge Graph and then proves the abstraction claim. Two Knowledge Graphs that look almost the same sit on two completely different databases, and the same question produces one identical VQL query in both.
It also shows where the numbers come from. Figures inside Veezoo's generated text reveal their source on hover, together with title and filters, so they always come from your data and never from the model alone.
Key points
- The structure of the Knowledge Graph is independent of the structure of the database.
- The same graph can sit on a normalized model or on a single wide denormalized table.
- Both graphs produced the exact same VQL query. Only the generated SQL differed.
Check your understanding
The same question ran against two almost identical Knowledge Graphs. What was the outcome?
Both graphs answered with 94 from the very same VQL query, while only the generated SQL differed.
What can differ between two Knowledge Graphs that look the same?
VQL is written against the graph, so identical graphs give identical VQL, while the deterministic SQL adapts to each database.
Your company migrates its data warehouse from Databricks to Snowflake. What happens?
The Knowledge Graph abstracts the physical model, so a migration only means remapping concepts to the new tables.