A zero-downtime migration affecting 70+ applications. 

How Railpen replaced an end-of-life data feed under a fixed cutover deadline and modernised its ingestion layer.

Summary

With £34 billion in assets under management and more than 350,000 members, Railpen safeguards the pensions of rail workers across the UK. Railpen needed to replace a core data feed that powered more than 70 critical business applications without disrupting any of them. With a fixed cutover window and an end‑of‑life dependency underpinning investment operations, reporting, compliance and day‑to‑day processes, the business could not tolerate downtime or data inconsistency.

We delivered exactly what Railpen required: a zero‑downtime migration, completed on time, with 100% data consistency verified across every consuming application. For the teams relying on this data, the transition was deliberately uneventful, the systems simply kept running.

Our approach, included a full data‑estate analysis, staged reconciliation, parallel running, and a continuously tested rollback path to ensure the migration was safe, predictable, and verifiably correct before cutover. The reconciliation tooling developed now provides Railpen with reusable assets for future platform changes.
Beyond addressing an immediate operational risk, the work has left Railpen with a stronger, more resilient and more governable data environment, forming the basis for continued collaboration on data‑focused initiatives.

Challenge

The existing solution that underpinned Railpen’s data ingestion reached end-of-life, setting a firm cutover date for evolution of their data platform. The data feed being replaced connected to Railpen’s on-premises infrastructure via SQL Server stored procedures — an extensive set of business logic, accumulated over many years, that was deeply embedded and hard to change. A new solution would replace it, and everything built on top of it needed to move across cleanly. 

The challenge was the scale of what depended on that feed. More than 70 critical business applications consumed data from it — covering investment data, reporting, compliance, and operational processes. Any failure to migrate would not be contained: the downstream impact would have been multiplicative across all of those systems simultaneously. The safe path required a thorough analysis of the data estate before a single line of new code was written. 

Approach

Before any build began, Softwire conducted a thorough analysis of Railpen’s data estate — mapping every data flow, every integration point, and every downstream dependency across the 70+ connected applications. This analytical foundation was the mechanism by which the team identified the migration path that would minimise downstream disruption and de-risk the cutover. 

Softwire, the platform supplier and Railpen’s technical teams operated as a single programme. The three organisations ran joint ceremonies — planning sessions, stand-ups, and triage — throughout the project, ensuring issues were surfaced and resolved collectively. There was no handover point; there was shared ownership of the outcome. This type of collaboration is what can make or break such a complex project and was one of the key drivers in the project’s success. 

• Analysed before building — A complete audit of the data estate established every integration point and the sequencing needed to maintain data consistency throughout the migration, not just at the point of cutover.
• Ran in parallel — The new platform ran alongside the existing feed, with reconciliation checks at each stage. The cutover only happened once the parallel run had demonstrated complete consistency across every application end-to-end. 
• Maintained a rollback plan — A tested rollback to the existing system was available throughout. The decision to cut over was made only when the team was confident it would not be needed. 
• Advanced the platform deliberately — Rather than replicating the stored procedure architecture, Softwire implemented Azure Functions — Railpen’s stated strategic platform direction. This was an opportunity to do things properly, not just to move things across. 
• Focused on data consistency at ingestion — Ensuring identical outputs at the point of ingestion meant downstream application teams could validate against a known baseline, containing the regression testing effort and keeping the programme on schedule. 
• Transferred knowledge throughout — Ongoing sessions kept Railpen’s engineers informed and capable. Ownership of the new platform passed cleanly to Railpen’s teams at go-live. 

Solution

In technical terms, Softwire replaced a connection built on SQL Server stored procedures — reading from an end-of-life data feed into Railpen’s on-premises infrastructure — with Azure Functions pipelines reading from a new Platform. The destination, Railpen’s on-premises databases, did not change. What changed was the ingestion layer: the pipelines, the processing, and the business logic that moves data from the platform into Railpen’s systems. 

This was not a lift-and-shift of the stored procedure logic. Significant business logic and data transformations that had previously been embedded across thousands of lines of SQL — difficult to read, difficult to change, and invisible to governance — were moved into the new platform, where they are visible, versioned, and fully auditable. The result is a data layer with end-to-end lineage that the previous solution did not provide. 

Key elements of what Softwire designed, built, and delivered: 

• Data estate analysis — Full mapping of all 70+ integration points, dependencies, and data flows prior to any build, establishing the migration sequencing and the risk profile for each application. 

• Azure Functions-based pipelines — Modern, cloud-native data pipelines replacing the SQL Server stored procedures, integrated into Railpen’s wider application landscape and aligned to its long-term strategic architecture. 

• Enhanced business logic transparency & control — Data transformations previously embedded in stored procedures were migrated to the platform, providing full lineage and governance over processing that had previously been opaque. 

• Staged reconciliation — Reconciliation checks ran at each stage of the migration. Discrepancies were resolved jointly by Softwire and Railpen’s teams before the next stage proceeded. 

• Parallel run and rollback — The new pipelines ran alongside the existing feed until end-to-end consistency was confirmed across all applications. A tested rollback plan was maintained throughout. 

• Reusable reconciliation tooling — The reconciliation tests developed for this migration are designed to be reused for future platform changes, providing lasting value for Railpen beyond the project. 

• Improved retry and logging — The new pipelines include structured retry mechanisms and detailed logging, providing clear visibility into failure modes and a fast, predictable path to recovery. 

Across the 70+ applications migrated, some integrations carried particular weight — either because of the frequency with which they were used, the business processes they supported, or the consequences of any failure. With 70 interconnected systems depending on this data feed, a failure to migrate successfully would not have been contained: the downstream impact would have been multiplicative. 

Results

The end-of-life data feed was replaced on time, within the required cutover window, with zero downtime and no downstream systems impact across all 70+ connected applications. For every team and every system that consumed data from this feed, the transition was a non-event — which is exactly what a migration of this kind should be, and rarely is. 

Beyond the delivery, the programme produced a material step forward in the quality and resilience of Railpen’s data infrastructure: 

• Zero downtime, on time – Cutover completed within the required window without disruption. 

• 100% data consistency – Verified through parallel running and reconciliation prior to cutover. 

• Risk reduction – Unsupported end-of-life infrastructure was retired, reducing operational and compliance exposure. 

• Improved resilience and observability – Structured logging, retry mechanisms, and clearer failure modes improved supportability. 

• Enhanced Data Management capabilities – Embedded data lineage and governance processes, providing traceability that the previous solution did not offer. 

• Reusable migration tooling – Reconciliation frameworks can be reused for future platform changes. 

“Softwire’s support has strengthened the resilience of Railpen’s investment and functional teams, creating a more robust foundation that enhances our ability to analyse and incorporate new data, and positions us to benefit from innovation now and in the future”

Julian Fava – Director of Enterprise Architecture and Data Governance at Railpen  

Continuing the Work

The migration programme was the beginning of Railpen’s relationship with Softwire, not the end of it. Following go-live, Railpen engaged Softwire on a subsequent project: data analysis work to support Railpen’s internal teams on a new data integration initiative. That follow-on engagement reflects the confidence built through the migration — in Softwire’s understanding of Railpen’s data environment, and in the way the two organisations work together.