Case Study: CHG Healthcare cuts a MySQL-to-Snowflake migration to 3 months with Datafold
Key results
The challenge
CHG Healthcare needed to translate more than 1,000 custom MySQL queries to Snowflake, many containing complex joins, subqueries, and non-deterministic behaviors. Reports had to produce identical results to maintain data parity, and manual validation of every query would have been time-consuming and error-prone.
The solution
CHG Healthcare used Datafold's AI-powered Data Migration Agent, which combined LLM-based SQL translation with validation-driven feedback loops, cross-database value-level diffing, specialized handling for MySQL-specific edge cases, and batched delivery via GitHub pull requests for CI/CD integration.
“The Datafold process — translating with the right LLM, validating the results, and letting humans review what falls out — is how I'd recommend every migration be done.”
AGAbigail GreenData Engineering Manager, CHG Healthcare
The results, in context
The migration was completed in 3 months against an estimated 18-month manual timeline, saving 15 months of work. Datafold reported the approach ran roughly 6x faster and about 80% cheaper than the alternative while preserving data parity.