Testing: The Bug Problem
In a recent post on ProjectManagement.com, I noted that there appears to be something very wrong with IT’s ability to deploy functionally correct and reliable applications despite the rigorous testing processes being followed today. Just this year alone, I witnessed two $20 million-plus deployments that were so defect riddled that neither produced the core outputs needed to be remotely useable.
In conversations with a number of my IT colleagues and peers about this issue, most echoed my experiences. What is going on? Why are so many defective core business applications being deployed?
In pondering this question, I first looked to see if there were anything these failed implementations had in common. Here is what I found:
- Development was outsourced to major providers.
- Communications between business SMEs and developers was sparse and cryptic at best.
- Data structures were overly complex and not properly normalized in many instances.
- Programs often contained literals that should have been parameterized as part of the data structure, often requiring major reconstruction efforts to resolve.
- On paper, testing methods and protocols were top notch, as was release management.
- Applications were very large, extremely intricate and complex with hundreds of interface and integration points to other systems (some legacy, some needing major improvement, etc.).
Please log in or sign up below to read the rest of the article.