How Cepheus helped a global insurance provider retire a legacy SAS and IBM/S370 mainframe environment, migrating to Azure Databricks with deterministic, audit-ready outputs.
The Client
A leading global insurance provider offering property/casualty, life and health, business insurance, and specialized lines such as trade credit insurance. The company had operated on a legacy SAS and IBM/S370 mainframe environment for over three decades, with these platforms underpinning underwriting, claims processing, actuarial modeling, and regulatory reporting.
The Challenge
- A mainframe-era foundation. Three decades of SAS and IBM/S370 infrastructure supported mission-critical insurance functions, but limited scalability, maintainability, and innovation.
- Non-deterministic outputs. Legacy SAS jobs produced inconsistent results due to missing unique keys, undermining trust in downstream reporting.
- No visibility into the estate. No inventory existed of SAS jobs, extracts, or reports — a significant blind spot for scoping and regulatory assurance.
- Encoding incompatibility. Legacy input files in EBCDIC format blocked ingestion into modern cloud pipelines.
- Thin SME coverage. Business continuity had to be preserved despite limited availability of domain experts who understood the legacy logic.
The Cepheus Approach
Cepheus didn't just convert code — we made three decades of legacy logic fully transparent and provably reliable.
- Full-estate lineage mapping to identify every dataset, report, and workflow dependency across the SAS environment
- Construct-level complexity analysis of macros, PROC steps, and data step logic to scope the true migration effort
- Engineered resolution of non-deterministic sorting to guarantee consistent, repeatable outputs
- Standardization of legacy EBCDIC files into ASCII for seamless ingestion into Databricks
- Semi-automated reconciliation for field-level validation and regression testing
- A structured four-phase delivery model: Scoping → Conversion → Testing → Hypercare
Major Challenges
- Non-deterministic sorting caused by missing unique keys in legacy SAS jobs
- Legacy EBCDIC-encoded files incompatible with modern cloud ingestion pipelines
- No existing visibility into the size or structure of the SAS workflow estate
- Complex, irregular SAS syntax requiring correction before conversion could proceed
- Limited availability of domain SMEs to validate legacy business logic
These are representative of the complexity Cepheus resolved — the full migration covered the entire SAS estate, with complete reconciliation and deterministic, audit-ready outputs.
The Impact
- 25% reduction in operational costs by eliminating SAS licensing and mainframe overhead
- 40% faster time-to-market for analytics workloads through cloud-native parallelization
- 100% deterministic outputs achieved through engineered sorting logic
- Full visibility into 30+ years of legacy workflows, enabling accurate scoping and planning
- Business continuity preserved despite limited SME availability
Why Cepheus
- True code translation — not stubs or placeholders
- Deep language engineering expertise to preserve and enhance business logic through conversion
- Construct-level SAS code analysis for accurate complexity estimation upfront
- Fully air-gapped, secure, and transparent toolchain with no proprietary runtime dependencies
- Commitment to full-estate migration, including tightly coupled and highly complex scripts
Still Running Core Insurance Functions on Legacy SAS or Mainframe?
If decades of undocumented logic are standing between you and a modern, cloud-native platform, we can help you migrate with confidence — not guesswork.
Talk to Cepheus