17
Importance of Integration
the new balance in the bank accounting records. However, if there is a
problem with the printing of a receipt (i.e., the written record of the account
number, date, time, and location of the deposit), then there may be low coupling between the event of the deposition transaction and the written confirmation. Cohesion is the characterization of the measure of binding
between two objects through their interaction(s). The strength of interaction means that two objects can be coupled under various conditions in
which their interactions can change each other, whereas tight coupling
presents as the observed causality between two objects. Loose coupling
implies either that many variables are at work and therefore observations
show weak causality between actions of the two objects, or that the observations do not reveal the linkage(s) between the two objects (although the
connectivity and the relation between the two objects are represented or
known). Coupling is a measure of the relation between objects. Unlike coupling which is determinable by a relation-in-fact through connection(s) that
are direct and causal, cohesion is the consequence of a relation-by-degree.
The relation-by-degree is observable, identifiable, and referenced to a particular object. Tight coupling can spawn dependencies that are distinguishable as causal. Tight coupling presumes a high degree of “trust” between
objects. In contrast, cohesion is the manner in which one object relates to
another object. Cohesion is formed by interactions of EMMI across boundaries of two or more objects. Some of these interactions are by process,
some by functions, some by behaviors, some coincidental, and some temporal. All these types of interactions can be summarized as cohesion by
structure and coupling by circumstance. Any manner and means of interaction may result in a form of cohesion. An operative definition of cohesion
is the minimum number of objects which, if removed from a group of
objects, would disconnect the interactions of the group. This definition is a
modification of that presented by Moody and White (2003). The many pertinent definitions of coupling and cohesion can be applied to systems integration as the overall point and expectation of integration are to bring
together objects to produce cooperative, unified system-level functionality
(Pikula and Siemion 2007).
In essence, the measurable concepts of connectivity, coupling, and cohesion are other ways of expressing interactions between objects. Coupling and
cohesion are defined as measurable concepts rather than specific measures
(Darcy et al. 2005). As such, coupling and cohesion are direct indicators of
interaction. If either coupling or cohesion is of small consequence, then two
objects are interacting and have met the threshold for interaction.
From an integration perspective, connectivity, coupling, and cohesion
are designed into the structure and action(s) of each object and exhibit
the characteristics expected by the design, concept of operations, and the
architecture. Integration testing confirms that the objects behave in the
manner expected by demonstration of their functionalities with various
performances.