Airlines
Wipro · LATAM
India · December 2025
Oracle to Azure PostgreSQL & AKS — DMAP AI-led reinsurance modernization, 30% faster with zero critical defects
A leading multinational life insurer engaged Newt Global to modernize a 20-year-old, Oracle-based Reinsurance Information System (RIS) to Azure Database for PostgreSQL Flexible Server and Azure Kubernetes Service (AKS). Using DMAP AI for assessment, schema conversion, embedded-SQL remediation, and data validation, Newt Global migrated 3.7 TB of production data and 210+ database-heavy batch programs — delivering a 30% performance gain over Oracle, six-decimal-place calculation parity, and approximately USD 150K in annual savings, with zero critical defects after go-live.
Oracle → PostgreSQL
AKS / Tomcat
Azure Key Vault
Terraform IaC
DMAP AI
Microsoft Azure
Customer
leading multinational life insurer
- Industry
- Primary Location
- Data volume
- Assessment
- Optimized DB size
- Performance
- Published
- Insurance
- South East Asia
- 3.7 TB
- 17 TB
- 3-week DMAP AI-led assessment
- +30% vs. Oracle
- February 2026
Production data migrated to Azure PostgreSQL
Performance improvement over the legacy Oracle platform
Annual licensing and infrastructure savings
Zero
Critical defects during the post-go-live warranty period
Background
The customer is a leading multinational life insurer whose Reinsurance Information System (RIS) had run for more than two decades as the engine for reinsurance treaty calculation and reporting across life, group life, and direct-marketing life product lines. The platform sat on a legacy on-premises stack — an Oracle 19c database and an IBM WebSphere application tier — that had become expensive to license, difficult to scale, and increasingly hard to support.
The insurer set out to modernize RIS onto Microsoft Azure: moving the database to Azure Database for PostgreSQL Flexible Server, re-platforming the application from WebSphere to Tomcat on Azure Kubernetes Service (AKS), and remediating two decades of embedded Oracle PL/SQL — all without disrupting business-critical premium and claim calculations. Newt Global was engaged as the implementation partner, with Microsoft supporting the Azure architecture.
The challenge
RIS is a calculation-critical system. Reinsurance settlements depend on results that must match the legacy Oracle output to the last decimal place — any drift is a financial and compliance risk. Modernizing it meant tackling several hard problems at once:
- A 20-year-old codebase with Oracle PL/SQL embedded across 544 Java files and numerous shell scripts, all requiring conversion to PostgreSQL-compatible logic.
- 210+ database-heavy batch programs driving premium and claim calculations, each of which had to produce identical results on PostgreSQL.
- 3.7 TB of production data spanning hundreds of tables, to be migrated and validated with zero tolerance for loss or corruption.
- A re-platform from IBM WebSphere to Tomcat on AKS, plus a full Azure landing zone, security model, and CI/CD pipeline built from scratch.
- Meeting a strict processing window: millions of premium records had to be processed within a fixed nightly/period-end timeframe, with performance at least on par with Oracle.
Key obstacles
- Calculation accuracy had to match Oracle to six decimal places — a demanding bar for actuarial reinsurance workloads.
- PostgreSQL and Oracle differ in ways that directly affect financial math: rounding behavior, index strategy, empty-string and NULL handling, date arithmetic, and package-global-variable semantics.
- Large-table performance at production scale (billions of rows) behaved very differently from smaller test datasets, surfacing bloat, index-depth, and query-plan issues only under real volume.
- The legacy system offered limited disaster-recovery guarantees, and the business needed a materially stronger resilience posture after cutover.
The solution
Newt Global delivered the modernization through its DMAP AI platform and engineering services, combining automation at scale with hands-on performance and remediation work.
DMAP AI-led assessment
A three-week, DMAP AI-driven assessment scanned the database schema and full application repository to quantify scope, automation feasibility, total cost of ownership, and one-time migration cost. It produced a detailed object inventory, embedded-query analysis, dependency mapping, and a fixed-price, milestone-driven execution roadmap with a clear RACI matrix.
Schema conversion & embedded-SQL remediation
DMAP AI automated schema conversion and the remediation of embedded Oracle SQL and PL/SQL into PL/pgSQL across Java and shell scripts — with side-by-side comparison, audit trails, and Git-integrated developer tooling. Objects with no one-to-one PostgreSQL equivalent were flagged with suggested fixes for targeted manual remediation.
Data migration at scale
The full 3.7 TB dataset was migrated and validated using a parallelized, month-wise approach for the largest tables, run across 8 parallel processing nodes and completed in roughly 32 hours. DMAP AI performed automated data validation — row counts, arithmetic validation of numeric columns, and BLOB/CLOB checks — to confirm parity between source and target before every sign-off.
WebSphere → Tomcat on AKS
The application tier was re-platformed from IBM WebSphere to Tomcat, containerized, and deployed to Azure Kubernetes Service with horizontal auto-scaling — replacing rigid legacy infrastructure with an elastic, cloud-native front end.
Secure, automated Azure foundation
Newt Global stood up a secure hub-and-spoke Azure landing zone provisioned entirely through Terraform infrastructure-as-code, with a Bastion-secured management hub, Azure Container Registry, and an Application Gateway as the single entry point. Credentials were centralized in Azure Key Vault, eliminating hardcoded secrets, and container images passed automated vulnerability scanning in the CI/CD pipeline.
Performance engineering
Newt Global's engineers resolved a series of Oracle-to-PostgreSQL performance differences to hit — and beat — the Oracle baseline. Examples included re-engineering index usage to eliminate DataFileRead waits (one job cut from 35 minutes to under 1 minute), replacing single large deletes with batched deletes (30 minutes to 3 minutes), tuning temporary-table statistics (4 hours to 90 minutes), adding pattern-optimized indexes for search (3 minutes to 20 seconds), and implementing a custom rounding algorithm to reproduce Oracle's commercial rounding to six decimals.
Resilient PostgreSQL architecture
The target ran on Azure Database for PostgreSQL Flexible Server with zone-redundant high availability in the primary region and a cross-region read replica for disaster recovery, backed by geo-redundant automated backups.
Migration journey
Phase 1 — Assessment & to-be architecture. DMAP AI one-click and detailed scans established scope, TCO, and a fixed-price SOW with a milestone roadmap and RACI.
Phase 2 — Schema conversion & remediation. Automated conversion of schema and embedded SQL, with targeted manual remediation for non-mappable objects.
Phase 3 — Data migration & validation. Parallelized, validated migration of 3.7 TB with automated parity checks between Oracle and PostgreSQL.
Phase 4 — App modernization on AKS. WebSphere-to-Tomcat containerization, Terraform landing zone, Key Vault, and secured CI/CD.
Phase 5 — Testing, parallel run & cutover. Functional, performance, HA, and DR testing, followed by an end-user parallel run comparing Oracle and PostgreSQL results before a controlled go-live.
The result
The reinsurance platform was successfully modernized from Oracle to Azure Database for PostgreSQL and AKS, delivering results the business had treated as aspirational at the outset:
- 30% performance improvement over the legacy Oracle platform — against an original target of merely matching Oracle.
- 19 million premium records processed within the required 10-hour window.
- Six-decimal-place calculation accuracy, matching Oracle exactly.
- ~USD 150K in annual savings on licensing and infrastructure — roughly 50% beyond the customer’s savings target.
- ~90% reduction in migration effort through DMAP AI automation.
- RPO reduced from 1 hour to zero, and RTO from 2 days to under 5 minutes for a regional failure.
- New environments provisioned in 30 minutes via Terraform, down from a 3-day SLA.
- 45% reduction in routine DBA effort by moving to managed Azure PaaS PostgreSQL.
- Zero critical defects during the three-month post-go-live warranty.
Key outcomes
- 3.7 TB migrated with validated data parity
- 30% performance gain over Oracle
- Six-decimal calculation accuracy preserved
- ~USD 150K annual savings (50% beyond target)
- ~90% migration effort reduction via DMAP AI
- RPO → 0, RTO → under 5 minutes
- Zero critical defects post go-live
About Client
A Leading Insurance company in South East Asia
About Newt Global Consulting
A leader in both heterogeneous and homogeneous database migration with application remediation. Newt’s DMAP platform delivers end-to-end, light-speed solutions for database, data, and application modernization on Google Cloud.
Ready to migrate?
See how DMAP can accelerate your Oracle-based Reinsurance Information System (RIS) to Azure Database for PostgreSQL Flexible Server and Azure Kubernetes Service (AKS). .
