Model Checking with Abstraction for Web Services

Abstract

Abstract. Web services are highly distributed programs and, thus, are prone to concurrency-related errors. Model checking is a powerful technique to identify flaws in concurrent systems. However, the existing model checkers have only very limited support for the programming languages and communication mechanisms used by typical implementations of web services. This chapter presents a formalization of communication semantics geared for web services, and an automated way to extract formal models from programs implementing web services for automatic formal analysis. The formal models are analyzed by means of a symbolic model checker that implements automatic abstraction refinement. Our implementation takes one or more PHP5 programs as input, and is able to verify joint properties of these programs running concurrently. IntroductionWeb services are instantiations of service-oriented architectures, where a service is a function that is well defined, self-contained and does not depend on the context or state of other services [1]. They are designed to be published, accessed and used via intranet or Internet. The elements of the design are (1) a service provider, which offers some service; (2) a service broker who maintains a catalog of available services; and (3) service requesters which seek for a service from the service broker, and then attach to the service provider by composing the offered services with its own components. A web service offers an arbitrary complex functionality, which is described in a global system structure. Examples of web services include information systems such as map or travel services, e-commerce systems such as web shops, travel agencies, stock brokers, etc. Clearly, it is essential to enforce security and safety requirements in the development of such systems.Web services are typically implemented in a very distributed manner and, thus, are prone to errors caused by the distribution. They often involve multiple parties. As an example, consider an online shop that accepts charges to a credit card as form of payment. The parties involved are the users or 122 N. Sharygina and D. Kröning customers, the vendor itself, and back-office service providers, e.g., the payment clearing service. The authorization of the payment is given by a service company for credit card transactions, whereas the "shopping basket" and warehousing are implemented by the vendor. It is easy to imagine another party involved in a transaction, e.g., a company that performs the actual shipment.Each party typically employs a large set of machines for the purpose of load sharing. The safety and security requirements are often global, i.e., require reasoning about multiple parties and may involve more than one server at each party.A typical scenario is depicted in Fig. 5.1. A merchant is operating three hosts (e.g., for redundancy or load-balancing reasons). Two of these hosts ('Host 1' and 'Host 2') are used to run a web server, e.g., Apache. The web server itself is split up into multiple processes, T 1 , . . . , T 4 . T...

Automatic Test Case Generation for Interacting Services

“…In this paper, we focused on business protocols in which data plays a secondary role. However, existing works such as [29,30] show that it is principally possible to add those data aspects into a formal model. As a result, we can refine not only the open net, but also the resulting operating guideline.…”

An Automated Method for Web Service Orchestration Based on Reusable Building Blocks

“…The work of Sharygina et al [31] does this to some extent, by verifying applications that implement web services and presenting techniques for modeling and verifying low-level languages used for implementation of web services (in this case PHP). However, this is restricted to verifying programs after their design is completed, while our approach enforces system correctness throughout construction by the use of ESMs.…”

Research on Web Service Composition Using Probabilistic Abstraction Refinement

“…The framework takes advantage of the characteristic behavior of a WS as a system that repeatedly takes a request and produces an appropriate response in order to simplify the abstraction. Sharygina N. and Kröning D. [25] gave a formalization of communication semantics for Web services, and an automated way to extract formal models from programs for automatic formal analysis. The formal models are analyzed by means of a symbolic model checker that implements automatic abstraction refinement.…”