FINNOFLEET

·

🇩🇪 Germany

·

500 employees

How FINNOFLEET Made Every Lending Decision Traceable with Event Sourcing

FINNOFLEET built its mars platform on Axon Framework to give German promotional banks a complete, auditable history of every promotional loan. Today it runs five products across around ten clients and several hundred production instances.

Introduction

FINNOFLEET Group develops specialized software for banks and financial service providers, helping them digitize, automate and efficiently manage credit and financial processes across the entire credit lifecycle. In Germany, it is one of the market leaders in software for promotional banking.

Its clients include state promotional institutions and the organizations that apply to them for funding, usually promotional loans. Many are intermediary institutions that handle promotional loan transactions with state promotional banks on behalf of their affiliated savings banks and cooperative banks.

In this market, one question has to be answerable at any time: who changed what, when, and why. FINNOFLEET answered it by making the history of every business transaction the system's primary source of truth. Since 2019, its mars architecture platform has run on event sourcing with Axon Framework at its core, giving clients a modern, modular platform that records every relevant business transaction completely and traceably.

The Challenge: An Audit Log Only as Good as the Code That Writes It

FINNOFLEET's legacy system was built on a classic CRUD architecture, where the audit log is derived from database operations rather than driving them. In a regulated finance and banking environment, that design left the team relying on a history that could fall out of step with the data it was meant to explain.

The business reason behind a change was lost

A CRUD update records the new value, not why it changed. For a lender that has to justify decisions to auditors and regulators, the missing business context is the part that matters most.

The audit trail depended on extra logic

Because the history was generated by separate code, it was only ever as complete as that code. Any gap or bug in the logging logic became a gap in the record.

Data and history could drift apart

The current data model and the history tables were maintained separately, so nothing structurally guaranteed they told the same story.

Modules were tightly coupled through a shared data model

Modules accessed each other's tables directly, so every module was bound to one shared data model.

The Solution: A Modular Monolith Built on Event Sourcing and Axon Framework

FINNOFLEET started with an architectural goal, not a framework. The team wanted a modular, event-driven architecture, then chose the tool that fit it best.

"It wasn't the choice of a framework that came first, but an architectural concept: we wanted a modular, event-driven architecture. The Axon Framework was the solution that best met our requirements."

Oliver Libutzki, Software Architect, FINNOFLEET

Why event sourcing: traceability comes first

Event sourcing stores every change to a system as an immutable event, and derives the current state from that sequence of events. For FINNOFLEET, that turned the audit trail from a by-product into the foundation of the platform.

"The most important factor was traceability. We operate in a regulated banking environment, so knowing who changed what, when, and why is a question that must be answerable at any time. With event sourcing, the history is no longer a by-product, but the primary data source."

Christian Raith, Compliance Officer, FINNOFLEET

The team saw three further benefits:

  • Loose coupling: Events are the natural contract between the modules of the modular monolith.

  • An expressive business model: Commands and events speak the language of the business, so business experts and developers work from the same model.

  • New views from existing data: New analyses and read models can be built from the existing event history, even for requirements nobody anticipated when the data was captured.

Why Axon Framework instead of an in-house build

FINNOFLEET first tried building its own event sourcing foundation. The team soon saw how long it would take to reach the maturity Axon Framework already had, and chose to spend that time on its own domain instead.

The mars platform architecture at a glance

  • Modular monolith: Business modules are clearly separated and communicate through events, but deploy together. FINNOFLEET gets decoupled without the operational overhead of distributed microservices.

  • Spring Boot and Axon Framework form the technical foundation.

  • JDBC event store: Events live in a relational database, a deliberate choice to avoid adding another infrastructure component.

  • Flexible operation: FINNOFLEET runs the applications as SaaS by default and can deploy them on-premises at a client's site on request.

  • Isolation by design: Each primary institution gets its own application instance with its own persistence layer, which keeps institutions as isolated from one another as possible and explains the several hundred production instances.

In the old system, history was derived from the state. On mars, state is derived from events, and the events are the record.

"In the world of event sourcing, this problem has been structurally eliminated: the state is derived from the events, not the other way around."

Oliver Libutzki, Software Architect, FINNOFLEET

Results: One Platform, Five Products, Several Hundred Instances

The mars platform has proven itself well beyond a single project. Five products run on it in production for around ten clients, across several hundred application instances, built by 20 to 30 developers following one standardized architectural model.

Steady delivery

FINNOFLEET ships a major update every six months, with smaller feature and bug-fix releases in between. Feature previews let clients test new functions early.

Isolation for every institution

Separate instances and separate persistence for each primary institution keep institutions as isolated from one another as possible.

Faster fault analysis

Some of the biggest benefits only became clear in operation. The event history has proven exceptionally valuable for analyzing errors: instead of laboriously reconstructing a state, the team can see which events, in what order, led to it.

Old cases stay traceable, new questions get answers

Because events are retained (anonymized where required), past cases remain easy to trace. And because new read models can be built from the existing history, new requirements apply to data the platform already holds, not only to data captured from now on.

Compliance built into the history: DORA, GDPR, and retention

FINNOFLEET's clients are subject to DORA, the GDPR, and statutory retention periods, so compliance plays a very large role in how the platform is built. The complete event history supports it in three ways.

  • Audit and history: Every business change is recorded as an event with its business context. The history comes first and is never reconstructed from technical operations.

  • Fault analysis: The same history used for audits makes operational faults easier to trace.

  • GDPR and retention in one model: Anonymization resolves the apparent conflict between an immutable history and erasure obligations. Personal data is anonymized while the business history is preserved, and because archiving is triggered by business events, the anonymization itself becomes part of the traceable record.

Building with the Axoniq Community

FINNOFLEET contributes to Axon Framework as well as building on it. The team is very active in the community, contributes through issues and pull requests, and keeps a lively exchange going with the developers at Axoniq.

FINNOFLEET’s developers are proud that the platform has proven itself across five products and around ten clients, and that they built their own concepts, such as anonymization, on top of Axon Framework, tailored precisely to the regulated funding sector.

Looking Ahead: Axon Framework 5

The organization sees real promise in Axon Framework 5 features such as Dynamic Consistency Boundaries (DCB) and the Postgres event store. Because the team built its own concepts, including anonymization, closely on top of Axon Framework 4, FINNOFLEET is taking a methodical approach to the upgrade.

FINNOFLEET's Advice for Event-Sourced Systems Built to Last a Decade

For teams considering event sourcing for a system that still needs to run ten years from now, FINNOFLEET offers three lessons:

  1. Invest in the mindset shift early. The real hurdle is the change in thinking, not the technology. Give it time and support, especially when many developers work on the platform.

  2. Plan GDPR and data deletion from day one. Don't adopt established patterns like crypto shredding without scrutiny.

  3. Anticipate growing event volumes. What looks fast in year one won't necessarily stay fast after years of operation, so factor replay times and large data volumes into the architecture from the start.

"Ask yourselves whether you can really know, at the very moment an event is created, what will need to be deleted later on."

Christian Raith, Compliance Officer, FINNOFLEET

FINNOFLEET's Blueprint for Regulated Software

FINNOFLEET set out to answer "who changed what, when, and why" at any moment, and built a platform where that answer is the data itself. Five products, around ten clients, and several hundred instances later, the mars platform shows that a regulated lender's software can keep a complete history and still adapt to requirements nobody saw coming.

Building software that has to answer to auditors?

Talk to an Axoniq engineer

Explore the latest with Axon Framework 5