Timely Delivery of Efficient Reliable Database Migration

A Legacy Innovation Story

The engineering team proudly presented their plan to adopt new platform languages and tools, apply Agile practices, and employ modern architecture patterns. A shy, doubting Thomas from Analytics asked, “I didn’t see in the plan how and when we migrate and refactor the data.” The Engineering lead quickly dismissed him, saying, “Oh, that bit is easy. We just run a few scripts towards the end of the project.”

Most data migrations fail because they overrun their time and cost budgets due to the disproportionate focus on code rather than data. In this talk, we look at the problems and solutions for a complex migration at ERPrUS. ERPrUS's business viability depended on the success of a large-scale data migration that required high performance, reliability, and aggressive timelines.

Fortunately, ERPrUS discovered an academic who had worked on a similar problem. The remainder of the talk describes the novel software engineering techniques, including bulk data conversion, decision tables, functional validation of algorithms and independent IV&V, used to tackle the challenges of migrating ERPrUS and their customers' data.