Noninvasive Re-architecture of Legacy Systems

Applied climatology is a numerically grounded discipline defined by long-lived observational studies that produce enormous quantities of sundry data using large and sophisticated software systems. Typically, multiple participants have organically evolved these systems over time creating a situation in which critical data is being handled by inefficient, difficult to maintain, and poorly understood software.

This type of situation has commonly been addressed by performing either a complete system re-write or an invasive re-engineering of the existing system in situ. Each of these techniques has its own well-documented set of drawbacks in time, cost, and risk. They also possess a shared flaw – when work is complete, system development usually resumes as before, eventually resulting in the system returning to the initial conditions that motivated the work.

To address these flaws, we offer a novel alternative approach for system normalization - an approach focused on re-architecture rather than re- engineering, requiring neither the large capital investments of a re-write nor the risk accounting measures of re-engineering.

Our approach cleanly maps generic legacy systems to modularized workflow architecture, non-invasively normalizing existing system behavior and providing a clear structure for future development.

We demonstrate this technique by exemplifying its use in the course of our work on a legacy NOAA NCEI ingest system that produces mission critical, nationally consumed climatology data.