Past experience of implementing applications that required excessive support and problem-fixing is one of the factors that have led most organizations to formalize processes throughout the lifecycle of design, development, testing, and deployment. The gravest consequence of lack of control over the quality of software is unexpected costs of ownership, such as the repeated need to fix problems – from the high-level business point of view, these undermine considerably any advantage that end users perceive in having their clever new application.
The approach of testing throughout the development lifecycle is very important, as this addresses a number of factors that govern whether the application is successful, such as whether it meets functional and performance requirements, and whether it can co-exist in a stable manner within its implementation environment.
Although the mashup approach is appropriate when there is less focus on discrete functional requirements – rather, making the most of what is already available functionally, and the fact that it is known to have undergone testing already – there is no advantage in short-cutting other types of testing, or failing to consider aspects such as how users will be supported after implementation.
Butler Group was recently told of an example of the latter within one organization, where a newly-available, free-to-own, heavyweight enterprise search facility, which can search folders and any type of content, was installed on a test server with indexes to a highly useful set of enterprise information. The search facility quickly gained popularity among the limited user population to which it was made available, being a marked improvement upon the facilities formerly available. Although it was explicitly only made available on a beta test basis, the search facility came to be used heavily every day and so became relied upon.
However, its test server platform was not under the control of production management, and so when the server was, quite legitimately, powered down without any warning having been given, the users of the search facility didn’t know where to turn as no formalized support mechanism had been put in place. In such circumstances, chasing about for answers typically involves a number of users, reducing the time that each of them has to devote to other productive purposes.
Without being seen as a barrier to the type of highly productive applications that can rapidly be delivered via mashups, IT managers must ensure that such facilities are not allowed to circumvent testing and release procedures, and that they benefit from support mechanisms that are there to serve the common good. It would be disappointing to see mashups heralded as a great means of improving productivity and flexibility, only to be perceived in hindsight to be out of control. Organizations’ previous strenuous efforts to be rid of seat-of-the-pants development practices should not be cast aside lightly.
Source: OpinionWire by Butler Group (www.butlergroup.com)