13
Importance of Integration
Principle 2: The Principle of Partitioning
Partitioning of objects can create tractable problems to solve if and only if boundary
contiguity is achieved.
Integration success thrives on simplicity. Simplicity is often achieved
by decomposing a high-level concept that embodies a few high-level objects
into multiple low-level objects in a hierarchical fashion. The high-level
objects (i.e., intellectual or physical) set the limits for all of the object
boundaries, that is, no object within the hierarchy of objects will have a
boundary limit that exceeds that of its logical high-level object. An
additional constraint for these multiple lower-level objects is their individual subboundaries do not overlap or underlap each other’s boundaries, if
the objects are on the same level. Further, an individual object (regardless of its
level in the hierarchy) is distinguishable by its mechanism, that set of
actions that converts an input into an output. The results of an object’s
transformation of EMMI through the actions of its mechanism(s) are the
object’s contribution to the performance of itself as well as the larger aggregation of objects. Partitioning an object or a set of objects into more objects
can create more manageable work packages to build and integrate the
objects. The ease of integration is facilitated, if and only if the object’s
boundaries are contiguous in terms of adjoining physical structures,
enabled functions that do not overlap or underlap with other functions,
and with whom user behaviors are uniquely identifiable. The object or
objects that are partitioned at the top level must be uniquely distinguishable and must cover the complete domain of the higher level partition(s).
There should be nothing left out of this top-level partitioning that does not
extend to the boundaries of the system. In other words, every object needs
to fall into a partition and stay within the boundaries of the top-level object.
The three boundaries (i.e., physical, functional, and behavioral) of each
object are by themselves the maximum extent of the object’s presence. In
aggregation, the objects form the system and its boundaries. When decomposing an object to its component objects, the combined boundaries of the
components must extend to the top-level boundaries of the parent object.
This condition of contiguity ensures that all that was conceptualized at the
top level was indeed included, nothing more (i.e., overlap condition) and
nothing less (i.e., underlap condition). Figure 1.1 illustrates the overlap and
underlap conditions.
Integration success requires that partitioning be carried out according to
this principle. Overlapping or underlapping boundaries between objects creates shared control over object mechanisms (overlapping condition) or lack
of control over a portion of an object (underlapping condition) which is identified as causing problems during development and integration. These
conditions that portend integration problems are not normally found in
interfaces or interface specifications as both of these conditions are symptomatic rather than causal for such problems.