Debugging Software

When light beer burst on the scene a generation ago, the biggest marketing challenge was to convince consumers that brewers could actually extract calories while still maintaining acceptable flavor. OK, that was a long time ago and now we have a full range of caloric options for beer. But every marketer (and beer drinker) will remember “Less Filling, Tastes Great”.

We will be making some exciting product announcements next week that addresses a similar kind of tension—this one in the software defect resolution space. Typically if a tester, developer or support person wants to get started on diagnosing an application bug, they’ll do some screen scrapes and a log dump but what really accelerates defect resolution are source-level root cause diagnostics. With current profilers, monitors and probe products, the tradeoff between “there’s bug in your code and here’s some logs” and “you have a fatal exception and here is the offending thread and corresponding source code” has, up until now, been very difficult to reconcile.

That’s right, up until now. Given the experience we have working deep in Java code execution and the 100% visibility into program execution that our app recording technology provides, we have come up with some very clever ways to deliver both triage speed and root cause depth in the same debug UI. The quality equivalent of great tasting light beer.

Stay tuned for some truly electrifying advancements in software defect resolution. (Hint: Sept 27).

Given the business we are in, software quality is one of our core motivations. A recent posting on STP, the Software Test Professionals site (http://www.stpcon.com/Item/1027/?utm_source=Email&utm_medium=email&utm_content=081611-TQA&utm_campaign=NEWSLETTERS) posed the question: “Can You Really Test Quality into a Product?”. The point of the article, that “doing it right the first time” is really the only guarantee of quality, harkens back to the seminal Phil Crosby quality principles that we have talked about in previous postings.

When it comes to producing quality software, it’s hard to argue against setting (and freezing) requirements, communicating them perfectly to all constituencies and then following a well-defined and monitored development process. That’s what we all aspire to. The problem is, just like with airplanes, automobiles and pharmaceuticals, the real world of complexity, human foibles and change always intrudes. That’s why testing exists—to catch defects before they get baked into the final product.

Software testing is the safety net underneath what is hopefully sound process and practice and while no one likes bug fixing, if defects are found early enough in the cycle, properly documented and communicated with enough information to find and fix them quickly, then quality will improve. It’s not the “free” quality that Phil Crosby espoused, but with the tools and techniques that are now available to QA and Dev teams, it’s less expensive that it used to be.

Anyone who lives on the California arc of the Ring of Fire knows that there is no such thing as geologic stability. We sit on tectonic plates that inexorably move and change direction, mostly unseen and unfelt until the big one(s) hit.

In the world of software, news that CA bought iTKO and Compuware acquired Dynatrace reminds us that as changes continue to grind away in the software development lifecycle, earthquakes like these two acquisitions will inevitably occur. The traditional silos of development, test and deployment/operations have been colliding for years as agile, DevOps and Cloud play to the overall need to reduce the time and effort to produce new business solutions.

The fact that two predominantly APM vendors have dived headfirst into ALM starkly illustrates the need to provide seamlessly integrated tools and management across a rapidly changing software development process. Leverage and value are now moving from “after the fact” monitoring and issue triage to pre-deployment detection and resolution.

ALM products are now at a premium because they can be effective at any point of the software development lifecycle.