One area where the interests of business units and IT departments are strongly in alignment is in rectifying the problems caused again and again by development projects that run late, and/or fail to deliver on the requirements promised. Far too many IT managers' reputations have been damaged due to lack of insight into projects, suffering the fate of ships without a beacon for guidance in the black of night, their projects smashed and splintered on the unforeseen, rocky shores of coding and budget problems.
We all know the basic foundations for being able to manage projects - the right information about what is actually happening, a comparison with what was intended at this point in time, and having this knowledge available in time to do something about things not being as they should.
Most organisations have provided themselves with tools to manage project-level information, but would still benefit from providing insight on errors more broadly, to those who can easily take the action necessary. Again, as we all know, taking that action at as early a stage as possible can dramatically save costs.
Many may not know that Borland provides a tool (Gauntlet) with which a troublesome range of error types can be identified as code is developed.
Borland acquired its Gauntlet product when it bought the company of the same name early in 2006, and it spent the rest of that year building more features, and integrating Gauntlet with other products within its lifecycle quality management (LQM) stable, itself part of Borland's more general application lifecycle management (ALM) focus.
Gauntlet runs in the background, checking code by running automated testing and test builds as developers submit their code to the organisation's source management repository. Its architecture facilitates the plug-in of different analysers, each of which can focus on a discrete type of problem that may be present in code. This is one of Gauntlet's major strengths, as Borland can readily add to the current range of problems that it highlights (environment integration, performance, security, vulnerabilities, and license compliance) for example, it could be extended in future to validate code against development standards.
In typical development programs, developers work to achieve a certain level of code quality before considering many of the types of problem that Gauntlet identifies, but then the re-work costs mount as code that has already undergone unit testing is revised, as integration, performance, and other error types that are Gauntlet's focus, begin to appear. That re-work expense is the first level of cost that Gauntlet helps to eliminate, but perhaps of greater benefit is the management-level insight it provides from the same underlying information.
By being able to identify developers from the IDs with which they check code into source management, Gauntlet can analyse the problem types that tend to be introduced per-individual (highlighting training needs), such as who is submitting untested code. It also provides project-level metrics on matters such as test coverage of the code base, so that managers can identify the extent of work remaining.
Gauntlet provides data that can be used to enable developers to fix problems at an earlier stage, without additional testing effort. It can give managers insight on progress against project plans, enable better go-live decisions, and use actual historical experience as part of decision-making. The underlying tests and builds used can be shared between development and quality assurance teams, saving further cost, and better integrating these teams. Butler Group believes that Gauntlet fits well into Borland's advanced development intelligence portfolio, and could deliver benefits to many end-user organisations.
Butler Group is a ComputerWire sister company and part of the Datamonitor Group.