Information Technology Reference
In-Depth Information
ship holds, perhaps with an associated warning about what the situation is
expected to be following the next upgrade, so that users have time to respond
to impending changes that would otherwise cause them problems.
If new product features become available, the user first has to decide
whether they offer sucient benefits to merit a migration campaign. Such a
campaign will normally start with an assessment of whether the changes would
offer local usability improvements within the existing business processes, or
whether a change to the business processes themselves is going to be needed.
Consider, for example, a workflow management system. This might have its
user interface upgraded, replacing a coarse-grained web forms interface with
a more dynamic AJAX-basedinterface without making a significant difference
to the business process, but requiring changes to the interfaces of a number of
decision-support plug-ins. Or, on the other hand, dynamic work rescheduling
features might be added to the same service, resulting in significant changes
to the business processes because they make possible finer scale targeting of
premium services to critical markets.
Once a level of abstraction has been identified at which the proposed en-
hancement is no longer visible, this level can be used as an overarching in-
variant and the change becomes a refactoring exercise aiming to maintain this
target description. This analysis helps to identify the scope of the changes
with confidence, since the invariant elements essentially limit the degree to
which consequential changes can propagate in the design. There is clearly a
strong parallel here with the discussion of federation in chapter 11.
12.3.3 The Changing Views of a Stakeholder
One of the consequences of opting for the use of externally purchased prod-
ucts and services is that there is a shift in the local stakeholder concerns, and
this, over time, results in evolution of the viewpoints used to model the system
as a whole. Something that was previously a major concern for engineering
design may cease to be of interest at all if responsibility for some particular
set of mechanisms is outsourced.
Consider, for example, the change in emphasis if an organization decides
to buy in a trust management solution from one of the specialist suppliers. A
whole range of encryption key-management technologies for which local skills
had been needed in the past become simply background information required
for risk assessments of the use of the external supplier, and the servers and
plug-ins needed become just black boxes. However, in consequence, there is a
strong dependency on the external supplier, and, in some respects, a reduction
in agility. Accepting this is as much a business decision as a technical one.
