From a single colocated server to a redundant cloud, without losing the PCI auditor along the way.
I joined Cash On Go at the end of 2013 as a lateral hire. My background was traditional system administration, the kind of work done by hand on a command line, one server at a time, where you remembered the topology because you had typed it. The company ran a consumer fintech product on a single colocated server in a Tallinn office space.
The assignment, stated plainly, was to move the application to something redundant and multi-region, without interrupting the service or the ongoing PCI-DSS audit. The assignment as I found it was broader: to build a team and a working practice that could survive the move and then keep going.
The hardest part of migrating production infrastructure is not the migration. It is being the person who has to say, out loud, that the current setup is not good enough — while that setup is paying the salaries.
The first thing I did was write it down. A short internal document, maybe ten pages, describing what was running, where, and what would happen to each piece under several plausible failure modes.
…continues.
The work I am proudest of from those years is not the AWS architecture, though it held up. It is the handover. When I left in 2017, the team kept running the systems without interruption. That is the hidden test of a migration: whether the person who replaces you can tell what you did and why, from the code and the notes alone.
The shape of that work is what I bring to every engagement now.