Data Handling
Scope
This page describes data handling for the Lineage Advisor for Microsoft Purview connector. It covers Marketplace subscription data, customer configuration, metadata, lineage records, operational logs, and support data.
Lineage Advisor is intended for business and enterprise use. Customers remain responsible for selecting the connected systems, credentials, metadata scopes, and Microsoft Purview collections used with the connector.
Data Categories
| Category | Examples | Purpose |
|---|---|---|
| Account and tenant data | Name, email, tenant ID, user ID, organization name | Authentication, tenant mapping, support. |
| Marketplace subscription data | Subscription ID, offer ID, plan ID, purchaser tenant, status | Provisioning, entitlement, billing lifecycle. |
| Connector configuration | Purview endpoint, collection ID, Databricks workspace URL, workspace scopes | Connection setup and sync execution. |
| Metadata and lineage | Asset names, table and column metadata, jobs, notebooks, workflows, pipelines, and relationships | Publish enriched Databricks lineage into Microsoft Purview. |
| Operational telemetry | Sync status, validation results, error messages, timestamps, request IDs | Reliability, support, security monitoring. |
| Support communications | Tickets, emails, call notes, attachments provided by customer | Customer support and issue resolution. |
Customer Content
Lineage Advisor is designed to process governance metadata and lineage. It does not require table row values to publish lineage into Purview.
Customer-controlled configuration determines which Databricks workspaces, catalogs, schemas, and Purview collections are in scope.
Customers should avoid placing secrets, credentials, personal access tokens, private keys, regulated data, or unnecessary personal information in asset names, descriptions, tags, support attachments, or other metadata fields that may be processed by the connector.
The connector does not intentionally inspect source table row data for lineage publication. If a connected source exposes sensitive information through metadata fields, logs, names, comments, tags, descriptions, or user-provided support materials, that information may be processed according to the customer's configuration and instructions.
Storage And Retention
Default retention targets:
| Data | Default Retention Target |
|---|---|
| Marketplace subscription and billing lifecycle records | Retained while the subscription is active and as needed for legal, tax, and billing obligations. |
| Connector configuration | Retained while the connector is active or until deleted by an authorized administrator. |
| Metadata and lineage cache | Retained while needed to provide sync, comparison, and troubleshooting features. |
| Operational logs | Typically retained for a limited operational window, then deleted or aggregated. |
| Support tickets | Retained according to support and legal requirements. |
Customer agreements, regulatory requirements, and enterprise deployment configuration may override these defaults.
Production retention periods may vary by deployment architecture, customer contract, regulatory requirement, and support process. Customers can request applicable retention details during procurement or security review.
Security Controls
Lineage Advisor uses:
- TLS for data in transit.
- Encryption at rest for managed storage.
- Least-privilege access patterns.
- Tenant separation controls.
- Operational logging for security-relevant events.
- Access restriction for production support and operations.
- Secret rotation and removal processes.
These controls are designed to reduce risk but do not guarantee that any system is free from unauthorized access, interruption, or data loss.
Credential Handling
Customers should use scoped credentials or service principals with only the permissions required for the selected Purview and Databricks scopes. Credentials should be rotated regularly and revoked when no longer needed.
Secrets should not be included in support tickets, screenshots, logs, asset metadata, or documentation examples. If a secret is accidentally shared, the customer should rotate or revoke it immediately.
AI And Model Use
Lineage Advisor does not use customer metadata or lineage content to train third-party foundation models by default.
If optional AI-assisted features are enabled, the enabled feature, data flow, provider, retention behavior, and customer controls should be documented in the customer's agreement or configuration.
Deletion
An authorized customer administrator can request connector deactivation and deletion of connector configuration. Deletion of Marketplace, billing, security, and support records may be subject to legal or operational retention obligations.
Deletion from the connector does not automatically delete records already published into a customer's Microsoft Purview account or data retained independently by Microsoft Marketplace, Microsoft Entra ID, Databricks, or other customer-controlled systems.
Incident And Support Data
When investigating operational or security issues, Lineage Advisor may ask customers to provide run IDs, timestamps, tenant identifiers, subscription IDs, sanitized error messages, or screenshots. Customers should redact secrets, tokens, private keys, and unrelated personal information before sharing materials with support.
If Lineage Advisor becomes aware of a security incident affecting customer connector content, it will follow its incident response process and provide customer notice as required by applicable agreements and law.
Security Review
For enterprise review, also see: