Platform implementation, architecture consulting, managed services
About PTSB
Permanent TSB is one of Ireland's pillar banks, serving over one million retail and business customers nationwide. As a systemically important institution in the Irish financial system, PTSB processes millions of daily transactions while operating under European Banking Authority regulations and Central Bank of Ireland oversight. The bank's technology infrastructure, built over decades around mainframe systems and batch processing, faced increasing pressure to meet real-time payment regulations and customer expectations. Following system constraints during peak trading periods that highlighted the limitations of batch-oriented architectures, PTSB committed to fundamental architectural transformation rather than incremental upgrades.
Goals & context
PTSB faced a critical juncture when EU regulations demanded sub-2-second payment confirmations while their mainframe-dependent systems required up to 10 seconds to process transactions. During peak periods like Black Friday, the mainframe's capacity limitations had already caused significant issues, unable to handle the transaction volumes required. Additionally, the mainframe's operating schedule created availability gaps where certain payment types couldn't process during offline hours, making real-time 24/7 operations impossible.
The immediate catalyst came through Safewatch, their fraud detection system, which mandated Kafka integration as a technical requirement. This exposed a fundamental gap: PTSB had zero internal Kafka expertise. When the architecture team investigated further, they discovered multiple vendor applications already running isolated Kafka instances within the organization, creating unmanaged complexity. While the architects had long advocated for event-driven architecture, they lacked both the business case for funding and the technical knowledge for implementation. The Safewatch deadline finally provided budget justification, though with an extremely compressed timeline of just two months to production.
PTSB evaluated multiple vendors including Confluent but concluded they needed more than just a platform. They required a partner who would provide hands-on support throughout implementation, help build internal capabilities, and work within their organizational constraints. The bank's vision extended beyond this single compliance requirement: they aimed to establish a foundation for customer notifications, mainframe offloading, and eventually a complete event-driven architecture transformation. However, with no existing dedicated event streaming team and development teams unlikely to gain deep Kafka expertise, they needed a solution that combined enterprise governance with self-service capabilities.
Strategic approach
Unlike typical Kafka implementations that defer governance until complexity becomes unmanageable, PTSB and Axual started with governance architecture to prevent the fragmentation already evident in their four shadow Kafka deployments. This wasn't about adding control layers to Kafka; it was about making Kafka invisible to business users through functional self-service.
The sequencing deliberately exploited organizational dynamics: using the Safewatch compliance deadline to fund the platform infrastructure, then leveraging that foundation for broader transformation. Where other vendors would demand completed requirements and formalized project structures, Axual agreed to begin with incomplete information and work within PTSB's constraints.
The operating model rejected both extremes of vendor dependence and DIY complexity. PTSB wouldn't become Kafka experts, but they would maintain functional control through self-service governance. Axual would handle the technical complexity of running Kafka on-premises while gradually transferring operational knowledge, not expecting PTSB to ever manage Strimzi operators or debug consumer lag issues.
99.99%
uptime
<1
second payment latency
4
months to production
Key initiatives: Dual-track streaming implementation
Discovery & challenge
PTSB's initial assessment revealed multiple vendor applications already running embedded Kafka instances across the organization, creating shadow IT complexity without governance or oversight. This fragmentation meant critical payment systems would add yet another unmanaged streaming layer unless the bank established enterprise-wide governance. Additionally, without self-service capabilities, every new streaming use case would require specialized Kafka expertise they didn't possess.
Technical architecture
Axual deployed a dual-track approach addressing both immediate compliance needs and long-term governance requirements. The streaming layer runs on PTSB's Red Hat OpenShift platform using Strimzi operators, with the architecture deliberately segregating payment streams from future enterprise workloads while maintaining unified governance across all streaming infrastructure. This on-premises deployment maintains data sovereignty within Ireland while delivering cloud-like self-service capabilities.
The self-service governance layer enables functional teams to provision topics, manage schemas, and control access through a web interface. This enterprise platform runs parallel to the payments infrastructure, maintaining complete isolation while providing centralized governance that reduces IT bottlenecks across distributed teams.
Implementation reality
The four-month deployment encountered challenges typical of consolidating preexisting Kafka deployments: multiple vendors working in parallel, evolving requirements as PTSB discovered what streaming could actually deliver, and incomplete prerequisites from fragmented ownership. These issues are standard when migrating from shadow IT to centralized streaming infrastructure. Axual guided PTSB through each discovery, adjusting implementation as their understanding matured rather than forcing predetermined patterns that wouldn't fit their reality.
The payment platform went live processing SEPA payments, meeting regulatory deadlines despite the original two-month timeline extending to four months. Payment processing latency dropped from over 10 seconds to sub-second response times, achieving EU regulatory compliance.
Roadmap
The enterprise platform's real test comes with customer notifications, where PTSB's digital team must successfully self-provision topics without Kafka expertise. Success here validates the governance-first approach and opens the path to mainframe offloading. The critical challenge remains organizational: transitioning from treating Kafka as vendor-mandated infrastructure to establishing a dedicated streaming team that can drive enterprise-wide transformation. Without this evolution, PTSB risks creating another generation of shadow IT, just with event streams instead of databases.
Partnership evolution
PTSB chose Axual over Confluent for one reason: they needed a partner, not a platform vendor. While Confluent would have provided documentation and professional services, Axual committed to being PTSB's de facto streaming team until internal capabilities developed.
The relationship deliberately evolves from intensive support to managed services as PTSB's teams adopt self-service capabilities. This isn't about vendor lock-in but practical reality: most banks will never build deep Kafka expertise, making the quality of ongoing partnership more critical than initial platform features.
Results
- Payment latency: 10 seconds → <1 second
- Regulatory compliance: 100% SEPA transactions meeting EU requirements
- Time to production: 4 months for mission-critical payment infrastructure
- Platform availability: 99.99% uptime since launch
Closing thoughts
PTSB's transformation proves that legacy banks don't need years of preparation or perfect conditions to adopt event-driven architecture. They delivered mission-critical payment infrastructure in four months despite having zero Kafka expertise and no dedicated streaming team, while managing multiple vendor relationships and evolving requirements. The key was pragmatic choices: using compliance deadlines to secure funding, selecting a platform with governance built in rather than attempting DIY, and partnering with a vendor willing to adapt to organizational realities rather than demanding rigid prerequisites. Their journey from 10-second mainframe processing to sub-second streaming demonstrates that technical debt isn't a barrier to transformation when you combine managed platform expertise with self-service capabilities.
Further information & resources
{{tenbtn}}
Related case studies
Read use cases and success stories from our customers
A Finnish TSO moved from Confluent to Strimzi with zero downtime and no rewrites, sustaining sub-ms latency and 100% uptime across 80+ apps.
Energy Trading with Apache Kafka - Becoming Faster Than the Competition
See how Enexis uses Axual to combine Strimzi's open-source framework with enterprise features to transform smart metering with real-time data streaming.