Partner Well Architected Reviews
Databricks conducts Partner Well Architected Reviews (PWARs) to validate partner integrations. A validation is Databricks' assessment that a partner integration is built to the security, reliability, auditability, and compliance standards defined by the Partner Well Architected Framework. A validation covers the product integration scope as reviewed, based on the materials provided at the time of assessment. It is not a permanent or open-ended approval, and is subject to revalidation at Databricks' discretion.
Review types
The review and validation path depends on your partner type:
| Partner type | What it covers | Validation guide |
|---|---|---|
| Connected ISV | Products that integrate with Databricks (BI, AI, data integration, governance) | Integration validation |
| Built-On ISV | Products built on Databricks as the core platform | Built-On validation |
| Data Collaboration | Data products shared via Delta Sharing and the Databricks Marketplace | Delta Sharing validation |
| MCP | External MCP servers published to the Databricks Marketplace | MCP Marketplace validation |
What a validation is not
- It is not a product endorsement or a guarantee of fitness for a particular use case.
- It is not a commercial agreement, or a substitute for one.
- It is not a substitute for a customer's own due diligence: customers remain responsible for evaluating the partner product against their security, legal, and compliance requirements.
- It is not a warranty of the partner product, and does not make Databricks responsible for its security, performance, or outcomes.
- It is not permanent: it reflects the integration as reviewed and remains subject to the conditions below.
Revalidation
A validation reflects the integration as assessed at the time of review. Databricks may monitor validated integrations on an ongoing basis, and partners must notify Databricks and obtain revalidation before deploying material changes. Triggers for revalidation include, but are not limited to:
Product changes
Material changes to the integration, including but not limited to:
- Modifications to an existing integration mechanism, or introduction of a new one
- Access to additional or different APIs, data, or metadata
- Changes to what data or metadata is stored, where it is stored, or how long it is retained
- Changes to data residency, processing location, or sub-processors
- Changes to the authentication model
- Changes to the deployment or tenancy model
- Introduction of a new product surface that alters the integration's scope or behavior
If an integration drifts from its validated configuration, the drift triggers revalidation and may invalidate the existing validation until the integration has been re-reviewed.
Business changes
A change of control of the company that built the integration. Databricks will revalidate the solution upon a change of control (as determined by Databricks). The revalidation must be completed before the integration may be utilized following a change of control.
Databricks requirement changes
Databricks reserves the right to update its validation requirements and standards, and to require revalidation of existing integrations as those requirements evolve.
Governing terms
Validation is governed by the Brickbuilder Partner Network Terms and Conditions and the Databricks Acceptable Use Policy.