A Combinatorial Procurement Auction for QOS-Aware Web Services

A Combinatorial Procurement Auction for QOS-Aware Web Services Composition Megha Mohabey, Y. Narahari Computer Science and Automation, Indian Institute of Science, Bangalore Sudeep Mallik, P. Suresh, S. V. Subrahmanya Infosys Technologies Bangalore September 2007 1 E-Commerce Lab, CSA, IISc

Overview • Web Services Composition Problem • Motivation, Research Gap and Contributions • QWESC: A Tool for QOS-Aware Web Service Composition • Combinatorial Auction for Web Services Composition • Conclusions and Future Work 2 E-Commerce Lab, CSA, IISc

Web Services • Network-resident software services accessible via standardized protocols • Can be automatically exposed, discovered, and invoked by other applications • Business processes and application functionality are becoming available as web services. • Can be endogenous or exogenous • Available often as commercial web services from enterprise solution vendors and web services marketplaces. – Eg. www. strikeiron. com, www. amazon. com, www. google. com 3 E-Commerce Lab, CSA, IISc

Web Services Composition Problem • Types of web services: – Stand-alone Web Services - Represent a single activity in a business process – Composite Web Services - Represent more complex business logic often encompassing multiple business steps • Web Service Providers (WSPs) may offer web services with different Quality of Service (Qo. S) and service level agreement (SLA) guarantees. • Dynamic creation of business processes requires composing an appropriate set of web services that best suit the current need. 4 E-Commerce Lab, CSA, IISc

Problem Definition • To find the best mix of web services that achieves well defined cost and quality criteria for the execution of an end-to-end composite service as desired by a WSR • We propose an approach to get the optimal execution plan for a single execution path. A B C A Simple Statechart 5 E-Commerce Lab, CSA, IISc

Related Work • L. Zeng, B. Benatallah, A. H. Ngu, M. Dumas, J. Kalagnanam, and H. Chang, “Qosaware middleware for web services composition, ” IEEE Trans. Softw. Eng. , vol. 30, no. 5, pp. 311– 327, 2004. – Qo. S modeling and Qo. S aware web services composition • Y. Gao, B. Zhang, J. Na, L. Yang, Y. Dai, and Q. Gong, “Optimal selection of web services for composition based on interface-matching and weighted multistage graph, ” in PDCAT ’ 05: Proceedings of the Sixth International Conference on Parallel and Distributed Computing Applications and Technologies. Washington, DC, USA: IEEE Computer Society, 2005, pp. 336– 338. – Interface matching • G. Canfora, M. D. Penta, R. Esposito, and M. L. Villani, “An approach for qos-aware service composition based on genetic algorithms, ” in GECCO ’ 05: Proceedings of the 2005 conference on Genetic and evolutionary computation. New York, NY, USA: ACM Press, 2005, pp. 1069– 1075. – Genetic algorithm based approach 6 E-Commerce Lab, CSA, IISc

Research Gap and Contributions • Current approaches consider the problem of composition of stand-alone web services to form an end-to-end composite service. • We present a novel combinatorial auction approach to dynamic web services composition that considers both stand-alone and composite services forming an endto-end composite service. • We design a software tool, QWESC (Qo. S-Aware WEb Services Composer), for web services composition based on the proposed approach. 7 E-Commerce Lab, CSA, IISc

Motivating Scenarios for Combinatorial Auction Approach – A WSP may be willing to provide a composite service at a price lower than the total price of stand-alone services that form the composite service. – The quality attributes of a composite service offered by a WSP may not be the same as achieved by composing its constituent stand-alone services. – Different WSPs may have different number of stages/steps for achieving the same task and they can expose the services at different points. 8 E-Commerce Lab, CSA, IISc

QWESC (Qo. S-Aware WEb Services Composer) Publish (WSDL) Enhanced UDDI Qo. S Registry Qo. S Ranking y er u Q se n po WSP 1 WSP n Service Template Qo. S Feedback s Service Re Description Discovery Engine Ma Web Service Ad tch ve ing Message Exchange rti Se se (SOAP) m rvic en ts e Web Service Preprocessor Binder and Abstract Process Designer Execution Engine Optimization Problem Qo. S Monitoring Optimal Formulator Web Service Solution Requestor IP Solver Qo. S Fee dba Payment ck E-Commerce Lab, CSA, IISc 9 Determination

Combinatorial Auction Approach • A web service requestor (WSR) wishes to buy a set of distinct web services A. • There is a set of web service providers (WSPs) who are interested in selling the entire set or some subsets of A. • The WSPs submit combinatorial bids • If a WSP can provide services A, B, C, AB, and BC, and there are two quality attributes, then the bid will look as follows: < ({A}, q 1, q 2, c(A)), ({B}, q 1, q 2, c(B)), ({C}, q 1, q 2, c(C)), ({A, B}, q 1, q 2, c(AB)), ({B, C}, q 1, q 2, c(BC)) > 10 E-Commerce Lab, CSA, IISc

Notation 11 E-Commerce Lab, CSA, IISc

IP Formulation 12 E-Commerce Lab, CSA, IISc

Experimental Results Announcements by WSPs 13 E-Commerce Lab, CSA, IISc

Experimental Results (contd…) 14 E-Commerce Lab, CSA, IISc

Experimental Results (contd…) THE VALUE OF VARIOUS ATTRIBUTES (C - WITH COMBINATORIAL BIDS, NC - WITHOUT COMBINATORIAL BIDS) 15 E-Commerce Lab, CSA, IISc

Future Work • Generalized Model – Current work limits discussion to that of a linear execution path. – Introducing forks and joins will model a more generalized scenario. • Interface Matching and SLA handling – The services offered by different WSPs may not be interoperable, if they have incompatible input output interfaces. – WSPs may provide service at various levels of quality guarantee violation of which may attract some form of penalty. • Game Theoretic Approach – A rational WSP may not find it in its best response to reveal true information about the web services provided by it. – To ensure truth revelation by WSPs, an appropriate incentive scheme has to be designed. 16 E-Commerce Lab, CSA, IISc

Questions and Answers … Thank You … 17 E-Commerce Lab, CSA, IISc