Information Technology Reference
In-Depth Information
to contain the quality dimensions (such as availability, exe-
cution time, price or throughput) identified and agreed by the service provider
and consumer. Each quality dimension has a domain and range; e.g., availabil-
ity is a probability usually expressed as a percentage in the range 0-100% and
execution time is in the domain of real numbers in the range 0
..
+
We define set
D
. A quality
dimension
d
can be considered
monotonic
(denoted by
d
+
)or
antitonic
(
d
−
);
monotonicity indicates that values closer to the upper bound of the range are
considered better, whilst with antitonic dimensions values closer to the lower
bound are considered better. A
parameter
m
associates a quality dimension to
a value range [3].
If a parameter is non-fuzzy (strict) its satisfaction degree will be evaluated in a
binary manner (Yes or No). In contrast, fuzzy parameters (relaxed) will be eval-
uated in a fuzzy manner which shows different degree of satisfaction (
x
∞
[0
,
1]).
Note that we also provide value ranges for both parameters regardless of being
fuzzy or non-fuzzy. The satisfaction degree of fuzzy parameters will be measured
using
membership functions
provided for each parameter. In the following we
provide the extended definition of a
parameter
based on the definition intro-
duced in [3]. In particular, we define a type
t
foraparameterthatcanbeeither
strict or fuzzy.
Definition 1 (Parameter).
We define a Parameter
m ∈M
as a tuple
m
:=
(
d, v, t
)
,d
∈
∈D
,v
∈V
,t
∈{
s, f
}
.where
D
is the set of quality dimensions,
V
is
the set of ranges for all quality dimensions
D
, s represent a strict parameter and
f represent a fuzzy parameter.
QoS once defined in a contract between two parties may change during a ser-
vice life-cycle. Changes could be due to system failures or evolution of quality
requirements from the involved parties. Therefore, Web Services need to be able
to adapt dynamically to respond to such changes. Adaptation and evolution
of services are playing an important task in this domain. However, adaptation
of web services needs to be performed in an appropriate manner to accommo-
date QoS changes/violations by choosing the best adaptation strategy. Defining
service description with the proposed fuzzy parameters provides a flexible situ-
ation in dealing with adaptation decisions. We discuss how it can facilitate the
adaptation of web services through an example. According to the new definition
of parameters, we consider availability and response time as fuzzy parameters.
Let us assume an example of a contract with initial value ranges of availability
between 80% to 90% and response time between 2 to 5 seconds. We use this
example throughout the paper.
We provided situations in which new QoS ranges could be still acceptable for
both parties according to the existing contract [3]. We introduced a compati-
bility mechanism that used parameter subtyping relation and Allen's Interval
Algebra [2] for comparing value ranges and their evolution. The provider and
requestor are compatible with each other according to the existing contract if
the QoS changes are in one of the acceptable situations. If the compatibility
is not provided, however it does not give any information about the degree of
satisfaction/dissatisfaction of the offered service. For example if the new range
