Professional AssociationsProfessional Associations: ACM Distinguished Scientist | ACM Distinguished Speaker | Association for the Advancement of Artificial Intelligence (AAAI)
Synthy (2004 - 2008)
An End to End Approach for Composition and Adaptation of Web Services, And Relevant to Other Component Technologies
Technical Lead: Biplav Srivastava
Collaborators over the years: Vikas Agarwal, Girish Chafle, Koustuv Dasgupta, Gautam Das, Sugata Ghoshal, Mangala Gowri, Neeran Karnik, Arun Kumar, Ashish Kundu, Anupam Mediratta, Sumit Mittal and Sougata Mukherjea
External Collaborators: Prashant Doshi, John Harney
Availability: Synthy is freely available within IBM as-is. It can be provided externally but only under a commercial engagement.
We here discuss:
What exactly is Synthy?
Synthy is a technology for semi-automatically composing SOA-compliant components such that the new component meets the desired functional and non-functional requirements and the resultant component can be flexibly executed. While the technology is demonstrated using web services, the technique generally applies to services in Services Oriented Architecture (SOA), and possibly other component representations.
Applications are developed today in an ad hoc manner resulting in poor reuse of software assets and longer time-to-delivery. Viewing software components as web services, the current solutions to web services composition based on business web services (using WSDL, BPEL, SOAP etc.) or semantic web services (using ontologies, goal-directed reasoning etc.) are both piecemeal and insufficient for building practical applications. We have developed the first integrated and practical solution to composition of web services end to end from specification to deployment by synergistically combining the strengths of the current approaches.
There are two key novelties we have introduced: (a) We distinguish between web service instances that get actually deployed and web service types which represent a group of web services that have the same functionality. (b) We compose in stages representing what components are needed for the composite component and how the components should be connected together to realize the needed functionality.
The technology covers how the components capabilities should be represented, how the components should be composed and how the compositions should be managed/ adapted so that the user has minimal disruption during execution as the environment changes.
How does Synthy work?
A development organization maintains a registry of software components (web services, in our case) that are semantically annotated, using ontology to model the domain. When a new function/service needs to be created, the developer formally specifies the requirements of that service using terms from that ontology. Our tool (which uses AI planning-based techniques) then looks for relevant components in the registry, and stitches them together in a BPEL workflow that delivers the requested function. The control flow of the workflow is generated based on specifications that will change slowly (like the specification of the composite service) while the data flow of the workflow is generated based on factors that are quite dynamic (e.g. Quality of Service). We can also adapt to changes in the world by creating multiple alternative workflows and intelligently switching between them based on the severity of the change.
What makes Synthy special?
Synthy is unique in that (a) it is effective, (b) it is intuitively appealing and (c) it tries to position the role of different alternatives tried to date.
Synthy is effective as we have built a large set of compositions for a wide variety of domains including the domain of a commercial telecom customer. The composition built is correct by design (theoretically provable) and the tool will find one if a composition exists. In technical terms, the technology is sound and complete. The design of Synthy is also efficient and we have shown in simulated experiments that it can build composition in domains with large number of components.
Synthy is intuitively appealing because it separates "what components need to be composed" from "how they will be composed". We humans do this in practice and this approach is also theoretically sound.
Synthy also brings out the synergy among the current approaches. The business world has adopted a distributed systems approach for composition in which the web service instances are described using Web Services Description Language (WSDL), and published to a Universal Description, Discovery and Integration (UDDI) directory, composed into flows in a language like Business Process Execution Language for Web Services BPEL), and invoked with Simple Object Access Protocol (SOAP). The academia has propounded the AI approach of formally representing web service capabilities in ontologies, and reasoning about their functional composition using goal-oriented inferencing techniques from planning. These approaches by themselves are piecemeal, and insufficient. The former has focused on the execution aspects of composite web services, without much consideration for requirements capture and the development process. The latter approach has stressed the feasibility of service composition based on semantic descriptions of service capabilities, but its output cannot be directly handed off to a runtime platform for deployment. We use both - the academia approach is used to answer the "what" of the composition while the industry approach is used to answer the "how".
IP:Papers and Patents
- The Synthy Approach for End to End Web Services Composition: Planning with Decoupled Causal and Resource Reasoning , B. Srivastava, NECTAR paper in Proceedings of the 21st National Conference on Artificial Intelligence (AAAI-06), Boston, USA. [Describes Synthy from a planning point of view.]
- Synthy: A System for End to End Composition of Web Services , V. Agarwal, G. Chafle, K. Dasgupta, N. Karnik, A. Kumar, S. Mittal, B. Srivastava, Journal of Web Semantics, Vol. 3, Issue 4, 2005. [A Comprehensive description of Synthy.]
- A Service Creation Environment based on End to End Composition of Web Services , V. Agarwal, K. Dasgupta, N. Karnik, A. Kumar, A. Kundu, S. Mittal, B. Srivastava, Proceedings of the 14th. WWW Conference (WWW 2005), Japan. [Introduces the Synthy idea. Suggested as the first read.]
- An Integrated Development Environment for Web Service Composition , G. Chafle, G. Das, K. Dasgupta, A. Kumar, S. Mittal, S. Mukherjea, B. Srivastava, In Proceedings of the IEEE International Conference on Web Services (ICWS 2007), Industrial Track, Salt Lake City, USA. Also as IBM Research Report RI 06009, November 2006. [Describes the Synthy tooling.]
- Improved Adaptation of Web Service Compositions Using Value of Changed Information , G. Chafle, P. Doshi, J. Harney, S. Mittal, B. Srivastava, Industrial Track paper in the Proceedings of the IEEE International Conference on Web Services (ICWS 2007), Salt Lake City, USA. [Describes improved mechanism for web services adaptation using the value of changed information in the Synthy system.]
- Adaptation in Web Services Composition and Execution , G. Chafle, K. Dasgupta, A. Kumar, S. Mittal, B. Srivastava, Industrial Track paper in the Proceedings of the IEEE International Conference on Web Services (ICWS 2006), Chicago, USA. [Describes mechanism for web services adaptation in the Synthy system.]
- Information Modeling for End to End Composition of Web Services , A. Kumar, S. Mittal, B. Srivastava, Proceedings of the 4th International Semantic Web Conference (ISWC 2005), Galway, Ireland. Also, IBM Research Report RI 06009, November 2006. [Discusses information modeling issues in Synthy with focus on types v/s instance, roles and handling of non-functional requirements.]
- Understanding Approaches for Web Service Composition and Execution , V. Agarwal, G. Chafle, S. Mittal, B. Srivastava, In ACM Compute 2008, Bangalore, India. Also in IBM Research Report RI 07005 [Formalizes the web services composition and execution problem, and explains which approach to use in a specific scenario. Successor to AAAI05 Workshop paper below.]
- Evaluating Planning based Approaches for End to End Composition and Execution of Web Services , V. Agarwal, G. Chafle, K. Dasgupta, S. Mittal, B. Srivastava, In AAAI 2005 Workshop on Exploring Planning and Scheduling for Web Services, Grid and Autonomic Computing, Pittsburgh, USA. [Characterizes the web services composition and execution problem, and explains Synthy and other planning-based approaches for it.]
- Applying Planning in Composition of Web Services with a User-Driven Contingent Planner , A. Mediratta, B. Srivastava, In IBM Research Report RI 06002, February 2006. Accepted as short paper in ICWS 2006 but we refused publication. [Explains the planning features in Synthy.]