Information Technology Reference
In-Depth Information
et al. [5] presented a re-composition approach dealing QoS replanning issues at
run time using late binding technique. However, none of these works consider
the consequences and potential overheads of adaptation. To this end, for exam-
ple, an environment for compensation of Web Service transactions is proposed
in [25]. In order to consider the overall value of a change, [15] presented an ap-
proach called
value of changed information
(VOC). Furthermore, an adaptation
mechanism is proposed based on VOC in [8]. However, these works have the
limitation that they do not take into account the satisfaction level of services.
Making adaptation decisions and evaluating them is therefore complicated and
has consequences that are often neglected. Some qualitative and quantitative
techniques has been proposed, however evaluating impacts of adaptation still
remains as an open challenge.
One core issue to address is the definition of a flexible description for Web Ser-
vices. Formulation of service specifications/requirements has been studied in the
literature. In autonomic systems and in particular web services, reasoning about
such specification is a hard job due to the changing environment that affects ser-
vice requirements. Although a lot of research has been conducted for functional
Web Service description, only a few efforts have been done with respect to non-
functional properties description of Web Service. Among syntactic and semantic
WS description we refer to the work done in [30], [21] and [16] which they also
provided algorithms for service selection based on such description. A major lim-
itation of those papers and other similar ones is due to not considering the partial
satisfaction of the QoS attributes. With this regards, [23] provided a semantic
Policy Centered Meta-model (PCM) approach for non-functional property de-
scription of web services. A number of operators (e.g. greaterEqual, atLeast) for
numeric values are defined in the model for determining tradeoffs between vari-
ous requests. Therefore, the approach can support the selection of Web Services
that partially satisfy user constraints. In [20] and [19], the authors extend the
approach proposed in [23] by proving a solution for Web Service evaluation based
on constraint satisfaction problem. The approach uses utility function to present
the level of preferences for each value ranges defined in the service description.
However, it does not take care of adaptation issues and controlling values at
run-time. In [22], the authors discuss about fixable and non-fixable properties
to deal with bounded uncertainty issue. Constraint programming is used as a
solution, however, there is no evaluation of the work.
It is required to provide a framework to evaluate alternatives and quantify
their impact for making decision decisions. Each alternative has different degree
of satisfaction and their impact has to be evaluated in order to select the best
adaptation strategy. A quantitative approach applying a probabilistic modeling
is used for partial goal satisfaction in [18]. Dealing with the uncertainty issue
is one major problem in order to formulate and manage service specification.
Thus, recently researchers are investigating to incorporate this uncertainty into
the service specification. In [14], the author provides support for reasoning about
uncertainty. A goal-base approach for requirement modeling in adaptive systems
is proposed in [9] which uncertainty of the environment is taken into account.
