SDF is SQL with
User Defined Types
SDF supports classifiers which expand on traditional SQL data types. This metadata prevents logic errors, adds additional syntax to support a tighter integration with your editor, and fast data quality guarantees.
Classifiers Made for Data Developers
Prevent Silent Errors from Slipping Through the Cracks.
Augment SQL's primitive types (e.g. integers and varchars) with higher level types like UserID and PhoneNumber. Use classifiers to prevent logic errors like incorrect joins, bad aggregations, and misuse.
Find Issues Instantly with Static Analysis
Proper typing provides guarantees during development. Classifiers in SDF are fast, flexible, and computed on every compile, so you can catch downstream errors instantly while writing a SQL statement.
Easy Integration with Gradual Adoption
SDF classifiers are metadata on SQL statements and can run anywhere your SQL runs. Compatible with all supported databases.
Define Business Logic as Code
Every classifier improves your codebase. Rich types bring your SQL transformations closer to every business stakeholder by making queries easier to understand.
Why Define My Own Types?
Easier Data Understanding
Add context during development. Describe the shape of objects, metrics, and rules to understand your data more easily.
Results You Can Trust
Types are evaluated fast with static analysis and are always correct. They don't hallucinate, and they don't require a network connection.
Safety
at Scale
at Scale
From 10 tables to 1,000,000, classifiers are designed for scale. Classifiers enrich your other systems and support governance, privacy, and cataloging goals.
Classifiers Propagate Automatically
Classifiers flow through your entire warehouse, through lineage aggregations, and functions. Or, configure rules to tell SDF where to stop propagation or reclassify based functional transformations.
Typically, <1% of tables need manual classifiers
Classifiers Propagate to the Other 99% of columns
Reclassification and propagagtion are user defined and flexible
