Project Proposal
ECE 598SV Fall 2008
Francesco Sorrentino
October 2, 2008
1. Problem Statement
To specify and verify the correct design of fault tolerant Web-Services. The challenge is to
design and implement a tool that is able to verify if a service can correctly work under fault
hypothesis, technically speaking the goal is to prove the observational equivalence between
the specification of the service design and the specification of fault-free system.
2. Motivation
Web services are revolutionizing the way that e-commerce, consumer software, and telecom-
munication applications are provided. They enable user interaction through web-browser,
nowadays one can look for and buy almost anything online using such applications. The
importance of such tools is indicated by the rapid growing development in the industry stan-
dards (e.g., SOAP, UDDI, WSDL, BPEL4WS) and technology (e.g., IBM’s Web services
Toolkit, Sun’s Open Net Environment, Microsoft’s .Net etc).
There are various issues that have to be address in order to develop successful web-services.
In particular, a fundamental question was the modeling and the analysis of their interactions.
In the last few years many works have be done on formal models to check interactions of com-
posite web services [1, 2, 4, 5, 7]. Recently, in [3, 6], the authors treated interesting questions
as: Does a given service satisfy a safety specification? Can a given service be substituted by
a another service in an arbitrary context?
This proposal intend to investigate the model checking of fault tolerance properties (safety
properties) in a web service.
3. Approach and Deliverables
This project should be organized in II phases. The I phase will be dedicated to the design and
analysis of the fault tolerance properties and the second phase will consist in the development
of a tool that can verify automatically the fault tolerant properties of a given web-service.
In particular for the first phase we would like to reduce the model checking of this problem
to a satisfiability problem. In the second phase, we would build a frontend that translate
automatically the model and the properties in a boolean formula, so that we can use the
YICES[8] SAT solver, to check it.
By the end of the project, we intend to model check automatically if a given web service is
fault tolerant to crash or failures of some of its peers.
1
