How Cepheus helped a large US financial services firm migrate mission-critical SAS workloads to AWS Databricks — without disrupting business-as-usual or requiring repeated rounds of user acceptance testing.
The Client
A leading US financial services firm operating across personal loans, auto finance, credit cards, insurance, and capital markets. The organization had prior migration experience, having previously moved from Greenplum to Snowflake — but its SAS-integrated-with-Snowflake environment presented a new class of challenge: workloads that were both business-critical and deeply embedded in day-to-day operations.
The Challenge
- Manual migration was too slow. Early efforts required multiple rounds of UAT and extensive retraining of business users on unfamiliar codebases.
- Zero tolerance for disruption. Business-as-usual had to continue uninterrupted throughout the transition — a hard constraint given the firm's regulated, multi-line-of-business operations.
- Deep technical debt. Over 200 SAS macros and 100,000+ lines of code, including complex PROC SQL queries with multi-table merges, needed to be migrated without introducing output discrepancies.
- Outdated infrastructure underneath it all. The production Databricks environment itself was running an unpatched, six-year-old Python version, constraining what could be modernized in a single pass.
The Cepheus Approach
Cepheus didn't just migrate code — we made the outputs provably trustworthy from day one.
- Automated reconciliation to retrieve and verify all required SAS tables in a single pass
- Semi-automated comparison scripts to validate outputs and eliminate repeated UAT cycles
- Refactoring of pass-through SQL connections from SAS macros into native Databricks SQL
- Transition from session-based SAS authentication to secure, cloud-native secrets management
- Enablement of S3-based workflows, with hands-on training for business users moving off SAS-driven exports
- Migration scripts engineered to run reliably within the client's legacy Python environment, while laying groundwork for future modernization
Major Challenges
- Pass-through SQL connections tightly coupled to Snowflake ODBC, creating brittle dependencies
- Session-based SAS authentication incompatible with modern, cloud-native security practices
- Business reliance on SAS-driven ad hoc CSV exports with no direct cloud storage equivalent
- A six-year-old, unpatched Python runtime constraining what the migration could modernize
- A codebase of 200+ macros and 100,000+ lines, with frequent large-scale table merges and long-form PROC SQL logic
These are representative of the complexity Cepheus resolved — the full migration covered the entire workload estate, with complete reconciliation and zero unresolved output discrepancies.
The Impact
- 30% reduction in operational costs by retiring SAS licensing and streamlining Snowflake dependencies
- 50% faster migration cycles through automated reconciliation that eliminated repeated UAT rounds
- 100% auditability and repeatability of migrated outputs
- Improved business user adoption through simplified, Databricks-native workflows
- A future-ready architecture positioned for modernization beyond legacy runtime constraints
Why Cepheus
- True code translation of SAS macros and PROC SQL into Spark — not a rebuild from scratch
- Deep expertise in authentication and secrets management for regulated, cloud-native environments
- Construct-level SAS code analysis for accurate scoping and complexity estimation before work begins
- Commitment to full-estate migration with minimal disruption to business-as-usual
Still Running Critical Financial Workloads on Legacy SAS?
If regulatory pressure, licensing costs, or aging infrastructure are holding your data stack back, we can help you modernize with confidence — not disruption.
Talk to Cepheus