Automated performance assessment for service-oriented middleware

Socially Enhanced Services Computing

Dustdar

2011

Abstract-This paper addresses one of the major problems of SOA software development: the lack of support for testing complex service-oriented systems. The research community has developed various means for checking individual Web services but has not come up with satisfactory solutions for testing systems that operate in service-based environments and, therefore, need realistic testbeds for evaluating their quality. We regard this as an unnecessary burden for SOA engineers. As a proposed solution for this issue, we present the Genesis2 testbed generator framework. Genesis2 supports engineers in modeling testbeds and programming their behavior. Out of these models it generates running instances of Web services, clients, registries, and other entities in order to emulate realistic SOA environments. By generating real testbeds, our approach assists engineers in performing runtime tests of their systems and particular focus has been put on the framework's extensibility to allow the emulation of arbitrarily complex environments. Furthermore, by exploiting the advantages of the Groovy language, Genesis2 provides an intuitive yet powerful scripting interface for testbed control.

Section: A Related Research On Soa Testingmentioning

confidence: 99%

Script-Based Generation of Dynamic Testbeds for SOA

Socially Enhanced Services Computing

Dustdar

2011

“…For instance, SOABench [27] provides sophisticated support for benchmarking of BPEL engines [28] via modeling experiments and generating service-based testbeds. It provides runtime control on test executions as well as mechanisms for test result evaluation.…”

Section: Related Researchmentioning

confidence: 99%

Automating the Generation of Web Service Testbeds Using AOP

2011 IEEE Ninth European Conference on Web Services

Dustdar

2011

Abstract-One of the key concepts of service-oriented computing is dynamic binding which favors on-demand integration of services into a running system. Companies can outsource tasks to external partners by using their services and are able to switch over to alternatives at runtime. However, this flexibility comes at a high cost because it complicates the testing process significantly. The major problem with external services is their restricted usability for testing purposes, due to costs or because policies forbid trial invocations. In this paper we present our approach to solve this issue by generating automatically testbeds which emulate external services. Based on previous work on testbed generation, we have developed a technique for intercepting Web service invocations in Java-based systems, generating emulated replicas at runtime, and redirecting the invocations transparently. We describe the concepts of our approach, its applicability, and limitations.

“…SOABench [14] and PUPPET [15], for instance, support the creation of mock-up services in order to test workflows. However, these prototypes are restricted to emulating non-functional properties (QoS) and cannot be enhanced with programmable behavior.…”

Section: Related Workmentioning

confidence: 99%

CAGE: Customizable Large-Scale SOA Testbeds in the Cloud

Service-Oriented Computing

Schall

Mietzner

et al. 2011

Abstract. Large-scale and complex distributed systems are increasingly implemented as SOAs. These comprise diverse types of components, e.g., Web services, registries, workflow engines, and services buses, that interact with each others to establish composite functionality. The drawback of this trend is that testing of complex SOAs becomes a challenging task. During the development phase, testers must verify the system's correct functionality, but often do not have access to adequate testbeds. In this paper, we present an approach for solving this issue. We combine the Genesis2 testbed generator, that emulates SOA environments, with Cafe, a framework for provisioning of component-based applications in the cloud. Our approach allows to model large-scale service-based testbed infrastructures, to specify their behavior, and to deploy these automatically in the cloud. As a result, testers can emulate required environments on-demand for evaluating SOAs at runtime.