Conversion by Iterative Clusters

Whatever the conversion plan — from code cleanup to changing the source language — Cornerstone applies an iterative process to minimize the overall time to project completion.

A conversion process is inherently both manual and automated. The effort is non-linear. The conversion of the first few code modules in a system will take consideration, learning and time. Subsequent modules will take progressively less time as Cornerstone learns the coding style, techniques, software versions, and nuances buried in the legacy system. Eventually, at the end of the project, all the system components can be converted in a minimum amount of time with very high quality.

Here is the quality-time relationship related to the project iterations.


In order to spread the milestones, Cornerstone splits the conversion effort into clusters of modules. The first cluster will be small, possibly just one or two modules. Subsequent clusters become exponentially larger:

  • Staring Cluster: 1-2 modules
  • Second Cluster: 10-20 modules
  • Third Cluster: 100-200 modules
  • Fourth Cluster: typically all of the system

As each cluster is analyzed, it is tested to ensure point-to-point correspondence in the output. All business rules are verified. Discrepancies are traced to their source: perhaps an issue with the G4 tools, a data problem, or the logic in the underlying business rules.