Oracle’s licensing model had always imposed limits on customization, the ability to leverage internal views and extensions freely, and cost predictability. Those limits were manageable for some time, but as IBK’s infrastructure roadmap pointed increasingly toward scalable, AI-capable systems, the gap between what they had and what they needed became harder to justify.

Moving to EDB PG AI gave IBK a clean path forward: more control over cost, more freedom for engineering teams, and a platform better aligned to the bank’s long-term AI strategy.

The case for EDB PG AI: Compatibility, cost, and a sovereign data foundation that extends into AI

On the development side, the decisive factor was Oracle syntactic compatibility. EDB PG AI’s high compatibility with Oracle meant IBK’s existing SQL, PL/SQL procedures, and packages could migrate with minimal modifications, reducing both development effort and risk on systems whose precision is paramount. EDB’s Migration Toolkit automated significant portions of that conversion, compressing the timeline considerably.

On the operations side, the evaluation came down to cost structure, support quality, and where the platform could go. EDB’s licensing economics were a dramatic improvement over Oracle, and the commercial technical support gave the team confidence that financial-grade stability was achievable on an open source foundation. In addition, EDB PG AI’s native support for extensions including pgvector gave IBK a sovereign path toward AI and vector operations—not as a future platform decision but as a capability already built into what they were currently deploying.

15 production systems, purpose-built architecture

IBK migrated 15 systems in total, spanning the full range of core financial workloads, from real-time customer-facing platforms to TB-scale risk analytics infrastructure. Two systems in particular define the scope of what EDB PG AI is handling:

• Win Class, IBK’s VIP customer service platform, runs on a full high-availability configuration to meet its zero-downtime, real-time responsiveness requirements.
• The Early Warning System processes TB-scale data for financial risk detection, running dedicated online and batch databases in parallel to handle both real-time queries and large-scale batch workloads without compromise.

All 15 systems are in stable production today.

An integrated monitoring environment built entirely in-house

With the migration in production, IBK built its own monitoring system—primary/standby status, replication lag, and failover history across all clusters, visible from a single interface. SQL performance monitoring runs on EDB PG AI extensions and custom PL/SQL: slow-query detection, execution plan analysis, the full picture. No commercial tool dependencies.

“Rather than depending on commercial tools, we’ve built an environment optimized for our internal requirements,” Park says, “which has given us greater operational autonomy.”

A sovereign data and AI foundation for what comes next

The migration produced measurable results on both fronts that matter in a financial institution: budget and operations.

Licensing cost reduction versus Oracle registered as a leadership-level win on the IT budget. EDB’s high-availability setup and IBK’s self-built monitoring system together accelerated incident response and raised the stability floor across the environment. EDB PG AI’s Oracle compatibility shortened the project timeline itself, a result the team hadn’t anticipated.

IBK’s plan is to keep expanding. New systems and existing ones approaching license-renewal cycles are both candidates for open source DBMS migration. The longer-term vision is a data infrastructure strategy in which EDB PG AI serves as the sovereign platform at the center, extending from core data management into AI and data utilization on infrastructure IBK owns and controls.