SDLP - SQL Data Lineage Platform

SDLP (SQL Data Lineage Platform) is an enterprise lineage system of record for Microsoft SQL Server estates, developed by Gudu Software and offered by Logan Data. SDLP computes column-level data lineage across your entire estate by parsing stored procedures, views, functions, triggers, and jobs, then stores every derived lineage edge as an auditable fact in a governed repository.

SDLP is built on a simple principle: nothing is silently dropped and nothing is guessed. Every statement the scanner cannot resolve is recorded in a reject ledger with a reason code, a severity, and the raw SQL -- so lineage facts plus rejects always account for 100% of the scanned estate.

1. Export - A read-only login exports catalog metadata and SQL module text into an export bundle. SDLP never reads table data.
2. Scan - Every procedure, view, trigger, and job is parsed into the lineage repository, with batch tracking and resumable checkpoints.
3. Validate - An independent validator, built as a separate implementation, recomputes the identity hash of every stored row and scores each batch from 0 to 100.
4. Consume - Explore lineage in the web catalog with interactive graphs and search, or deliver it downstream through the Export Center, CLI, or export API.

• Column-level lineage with lineage roles: DIRECT, TRANSFORM, AGGREGATE, DERIVED, FILTER, JOIN, CONDITION, and INFLUENCED_BY
• Reject ledger with machine-readable reason codes and severity levels for every unresolved statement
• Hash-verified evidence: each fact opens the exact SQL statement that produced it, checked live with SHA-256
• Baselines that freeze lineage "as accepted on date X" for attestation
• Append-only, hash-chained audit trail for classifications, exports, and recomputes
• PII impact analysis: stewards classify columns and SDLP shows where that data flows and how -- contains data, filtered by, joined with, or influenced by
• Eight export types in CSV, JSONL, or Parquet, including a lineage CSV built for ingestion into Microsoft Purview with stable fact IDs
• Seven roles, isolated workspaces, and OIDC sign-in

• Tamper-evident trail: every change is written to an append-only, hash-chained event log
• Independent verification: auditors can run the validator themselves without relying on the scanner
• Baselines: freeze accepted lineage and attest against it at any time

• Runs entirely inside your own environment. Requires only Java on Linux or Windows and needs no outbound internet access, making it suitable for isolated and air-gapped networks.
• Choose an embedded repository (no database server required) for evaluation, or a SQL Server repository for shared production use.
• Typical time from download to first scan with the embedded repository: about 15 minutes.
• Reference scan: 122 databases (about 29,000 analysis units and 450,000 column-level lineage facts) completed in under 2 hours on a 4-core server. Actual results vary by estate size and complexity.

Data governance teams, data stewards, auditors, and scan operators at banks, insurers, and other regulated organizations that need to demonstrate where their data comes from and how it moves -- including regulatory lineage programs such as BCBS 239.

Compatibility: Microsoft SQL Server estates and folders of SQL files. Support for further database platforms is on the product roadmap.

Microsoft, SQL Server, and Microsoft Purview are trademarks of the Microsoft group of companies.