dbt Semantic Layer vs tbls

The dbt Semantic Layer centralizes metric definitions on dbt models and handles joins so downstream tools and applications can use metrics consistently. MetricFlow generates SQL at query time from semantic model and metric configurations. Teams define those models and metrics in YAML as version-controlled, lineage-aware code. MetricFlow can optimize joins, rewrite queries to reduce compute and latency, and cache repeated queries. The product offers an API, the dbt MCP Server and integrations for BI tools, notebooks, spreadsheets, embedded analytics and AI systems. GraphQL and JDBC APIs and a Python SDK provide ways to query metrics and dimensions; Claude and ChatGPT can connect to governed metrics through the MCP server. Documented warehouse platforms include Snowflake, BigQuery, Databricks and Redshift. Defining and querying metrics requires a dbt Starter or Enterprise-tier account, and single-tenant accounts need account-representative setup and enablement. Starter includes 5,000 queried metrics per month; Enterprise and Enterprise+ each include 20,000. Starter costs $100 a month per user. The free Developer plan does not list the Semantic Layer.

tbls is a command-line tool for documenting databases and checking their schemas. It generates GitHub Flavored Markdown and can export schema information in formats including DOT, PlantUML, Mermaid, SVG, PNG, JPG, JSON, YAML, Excel, and configuration files. Its configurable lint rules can require column comments or set column-count limits. The `diff` command compares a database schema with generated documentation or another database, while `coverage` measures documentation for descriptions and comments. Documented data sources include PostgreSQL, MySQL, MariaDB, SQLite, BigQuery, Cloud Spanner, Amazon Redshift, SQL Server, Azure SQL, Microsoft Fabric Warehouse, DynamoDB, Snowflake, MongoDB, ClickHouse, Databricks, JSON, HTTP, and GitHub. Docker and a GitHub Action are documented for CI use, and external drivers can extend database support. It runs on Linux, macOS, and Windows, or in a self-hosted setup. The listed plan is free under the MIT License. SQL Server requires version 2017 or later and database compatibility level 110 or higher.