The client, a Canadian provincial health authority, had built its analytics capabilities on a SAS grid-based Business Intelligence Environment over many years. What had once been a functional foundation had grown into a complex, costly platform that required significant ongoing investment in infrastructure, database licensing, and tooling administration to keep running. The organization recognized both the operational burden and the strategic opportunity: modernizing onto Snowflake would reduce cost, simplify administration, and position the health authority’s analytics environment for the future.
The challenge was the scale and interconnectedness of what needed to move. The existing estate contained a large number of SAS objects, each with its own dependencies, that had to be discovered, sequenced, and migrated with full awareness of how they related to one another. Manual approaches to discovery and code conversion at this scale are slow and error-prone, with a high risk of introducing inconsistencies that compound as migration progresses. The client needed a structured, repeatable method that could handle the full complexity of the estate without disrupting the analytical work that depended on it.
IBM Consulting deployed an AI-assisted SAS Migration Assistant to systematically modernize the client’s analytics ingestion layer onto Snowflake, following a phased delivery model designed to provide real value at each stage of the migration journey rather than deferring all benefit to a distant endpoint.
The foundation of the approach is a knowledge graph that maps the full web of dependencies across the SAS estate, giving the client a structured, comprehensive view of how its analytics objects relate to one another. This dependency map drives the sequencing of migration work, ensuring objects move in the right order and that nothing is inadvertently broken in the process. From there, deterministic AI code conversion translates SAS code into Python in a consistent, repeatable way. By removing the variability of manual conversion, the approach produces results that engineering teams can review, validate, and trust. Applicable processing logic is then implemented directly in Snowflake, with engineering review and validation built into each step.
Delivery is structured in phases: the bronze ingestion layer is the first milestone, establishing the core data foundation on Snowflake. Silver and gold transformation layers, along with broader SAS enablement capabilities, follow in subsequent phases. This phased path lets the client begin realizing the operational benefits of the Snowflake platform now while the broader migration progresses on a clear, governed schedule.
