A data contract captures the promises a piece of data makes to the people who use it — who it's for, how fresh and reliable it is, what quality rules it must pass, and which systems it depends on. In dScribe, contracts are built on the Open Data Contract Standard (ODCS), an open, portable format rather than a dScribe-only definition.
By default, a data contract is attached to a data product, but it doesn't have to be — see contracts on other asset types below.
Versioning
Contracts are versioned. From the Contract tab you can see the active contract version, create a New version, roll back to a previous one, and view the generated YAML file.
What a contract defines
Lifecycle status & consumers
Set where the data is in its lifecycle (for example Production) and define its audience by adding consumer groups (for example Sales Viewers). Together these tell users whether the data is ready to rely on and who it's intended for.
SLAs
Service level agreements are the measurable commitments the data makes. Add an SLA property and give it a target value, unit, description, and an optional driver (the reason behind it, such as regulatory).
The available SLA properties are Availability, Latency, Throughput, Error Rate, Retention and Freshness:
Data quality rules
Quality rules are defined on the contract's datasets and on individual dataset elements (columns), and shown alongside the schema. For how rules are built — severity, rule types, metrics, thresholds, and external engines — see Data Quality.
Infrastructure
The Infrastructure section lists the source systems the data draws on (for example Fabric DWH and Power BI), so consumers can see at a glance where the data comes from.
Contracts on other asset types
Data contracts are tied to data products by default, but they can be activated on other asset types as well — a Dataset, for instance — when you want the same data contract guarantees on an individual dataset rather than a packaged data product. You can configure this via Asset Types.
Where to go next
→ Data Products — bundle assets into a curated product the contract backs
→ Data Quality — define, execute and follow-up on data quality rules via dScribe
Have a question or can't find what you're looking for? Use the chat icon inside the catalog to reach the dScribe support team.