Technical Overview & Strategic Context
When business metrics are defined separately in different BI tools, reports can display conflicting numbers. Data Fabric 2.0 addresses this by establishing a central semantic layer that stores metric calculations in code, exposing uniform definitions to all database engines.
Architectural Principle: Maintain metric and schema calculations inside source-controlled configuration files, not inside isolated dashboard scripts.
Core Concepts & Architectural Blueprint
Using tools like the dbt Semantic Layer, organizations write metrics definitions (like 'monthly active users' or 'net revenue') once. The semantic layer translates these configurations into SQL query parameters, ensuring consistency across dashboard tools.
Performance & Capability Comparison
| Metadata Layer | BI Tool Calculation Maps | Centralized Semantic Layer | Report Inconsistency Rate | |
|---|---|---|---|---|
| Metric Rules | Calculations defined in dashboard tools | Calculations defined in source-controlled yaml | High (metrics drift easily) | |
| Access Control | Rules configured per database instance | Centralized compliance and security rules | Zero metrics mismatch |
Implementation & Code Pattern
To write a metrics calculation definition in a dbt project, use this configuration template:
- ◆Write standard model schemas pointing to core database tables.
- ◆Configure metrics calculations in structured YAML configuration files.
- ◆Reference metric variables inside BI database connections.
# dbt semantic layer metric definition template (2025)
semantic_models:
- name: revenue_model
model: ref('fact_billing')
dimensions:
- name: payment_date
type: time
type_params:
time_granularity: day
measures:
- name: gross_amount
agg: sum
expr: transaction_amount
metrics:
- name: total_revenue
type: simple
type_params:
measure: gross_amountOperational Governance & Future Outlook
Centralizing business calculations inside a semantic layer ensures data consistency across reports, simplifies BI tool setup, and improves analytical accuracy.