A shared drive sat at the centre of the file-transfer estate. SAP pushed data into it. Other systems pulled data out. In between, undocumented shell scripts and cron jobs did whatever the original author had taught them to do years ago, on a different team, in a different version of the business.

It worked. That was the problem.

It worked well enough that nobody had to look at it. Until the day someone needed to change it.

Challenge

The hub had three problems that compounded each other.

It was a single point of failure. Every file flow in the estate routed through it. There was no fallback, no redundancy, no graceful degradation if it stopped.

It was undocumented. The shell scripts and cron jobs that handled the routing logic had accumulated over years. Nobody on the current team had written them. Nobody had full coverage of what each one did, in what order, with what dependencies.

It was unmonitored. There was no security perimeter, no audit logging, no central place to see what had moved through. If a flow failed, the failure was discovered downstream - usually by a business team that didn't receive what they expected.

The compounding effect was the real problem. You couldn't safely change the hub, because you didn't know what depended on what. You couldn't safely leave it alone, because the risk surface grew with every business change.

The "black box" wasn't a metaphor. It was an accurate description of how the operations team actually treated it: stay away unless something breaks, and even then, change as little as possible.

Solution

The shape of this project mattered as much as the engineering.

Two things had to happen, and they had to happen at the same time: the topology had to change, and the cron scripts had to be understood. Replacing the hub with another hub would have kept the black-box problem. Migrating the scripts as-is would have kept the unknowable logic. Doing only one would have failed.

So both happened in parallel.

The new design routes file flows directly between sources and destinations. The shared drive is gone. No central choke point. No single point of failure. Active Transfer Gateway sits in both internal and external zones, handling the secure transfer paths in a firewall-friendly, auditable way, but it doesn't store and forward. It moves.

The cron scripts couldn't simply be re-pointed. They had to be read first. Each script was traced and reverse-engineered to determine what it was actually doing - which sources, which destinations, which transformations, which schedules, which error conditions. Only then was the logic ported to the webMethods platform, where it now runs with monitoring and structured error handling.

Protocols moved up at the same time. FTP became SFTP and SMB. FTPS was retained where treasury required it. The migration was phased - flow by flow, business domain by business domain, validated against the old hub before the old hub stopped seeing the traffic. No data was lost in transit.

Results

The black box is gone. Every file flow is now traceable from source to destination, with no hidden intermediate hops and no undocumented logic in between.

500+ interfaces run on the modern platform - a fully transparent estate where every transfer is observable and every routing decision is governed.

Audit and compliance posture is restored. Monitoring and security policies are enforced at the platform level. The estate is ready for the kind of scrutiny the old hub couldn't survive.

Support is now possible. Maintainers can trace and troubleshoot using the platform's own tools. Nobody needs to learn legacy shell scripts or read a decade of undocumented cron jobs to make a change.

The architectural change is the easy part to describe. The harder change is harder to see: the team can now answer questions about the estate that the old hub made impossible to answer. What moved, when, where, why. That's a different kind of confidence to operate on.