International Journal of Progressive Sciences and Technologies (IJPSAT)
ISSN: 2509-0119.
© 2022 Scholar AI LLC.
Vol. 34 No. 2 September 2022, pp. 410-416
Corresponding Author: Pélagie HOUNGUE 410
Early Guessing of Performance Using Simulation as Part of
Service Development
Pélagie HOUNGUE, Romaric SAGBO
Information and Communication Technology department,
Institut de Mathématiques et de Sciences Physiques (IMSP/UAC)
Dangbo, Benin
pelagie.houngue@imsp-uac.org
Abstract— In Service-Oriented Architectures (SOA), the key problem is the quick and accurate evaluation of web service performance.
Despite the fact that the integration of the simulation step into the development cycle of softwares/web services can allow to learn early
the behavior of the performance of software/web service, it is still a challenge to use simulation as part of service development. This
integration can be used to assess the performance of a family of web services by developing one of them. In this paper, we propose a
methodology that shows how the simulation step can be integrated to the development cycle of a family of services using a model-based
approach to describe the services and by choosing a reference web service to be developed and used to guess the performance of the
remaining services in the family.
Keywords— Model-based testing; Simulation; Performance; Web services; Zero-knownledge
I. INTRODUCTION
The development based on model allows to integrate early a performance analysis at the software development stage [1], [2],
[3]. The evaluation of service performance is important [4], but in a Service-Oriented architecture there is some difficulties to
assess the non-functional properties of the web service without any knowledge on the service’s behavior. Nowadays, early
performance evaluation is a critical step in web service development [5] and it is important to have a prediction model for
performance evaluation at this stage of the development cycle when no historical data are not yet available [6].
In [5], we propose a model-based approach that uses a Symbolic Transition System (STS) to describe the web services as finite
state automata and evaluate their performance. This model was extended for simulation purposes by adding state transition
probabilities and the delay distributions on the transitions. We evaluated the accuracy of this approach using two different
scenarios, a full-knowledge and partial-knowledge scenarios which provide good results. Furthermore, in [6], we define a zero-
knowledge scenario which defines a model that allows to estimate the performance of the service when no a priori knowledge is
not available on its performance. This model uses the description of the service in the WSDL file, the SOAP message that
encodes the input parameters of the service and the STS-based model to guess the behavior of the service. Moreover, in [7] we
extends our previous works by using data obtained from our zero-knowledge approach, where neither the service code nor test-
based the information on service execution times are available, to generate the SLA template and use it to negotiate the
preliminary SLA and monitor it.
In order to show the application of our methodology to estimate early the behavior of the service, we propose in this paper the
different steps that need to be follow in order to use the simulation as part of the development [8]. This methodology will be
applied to a family of services and will show how the performance of the entire family can be guessed by choosing one of the