NOTATIONS
3.1 Introduction
Research on web service search and composition has become increasingly important in recent years due to the growing number of web services over the Internet and the chal- lenge of automating the process of composition. Since there is a large growth in num- ber of available web services and possible interactions among them are huge, searching for desired set of services to satisfy a user query becomes very difficult. This in-turn means service search and composition problem has become increasingly sophisticated
1Lakshmi.H.N, Hrushikesha Mohanty;Extended Service Registry to Support I/O parameter-Based Service Search,Proceedings of 1st International Conference on Intelligent Computing, Communication and Devices, ICCD 2014, Springer Publications,Volume 308 AISC, Issue Volume 1, 2015, Pages 145- 155.
2Lakshmi.H.N, Hrushikesha Mohanty;RDBMS for service repository and composition, Proceedings of 4th International Conference on Advanced Computing, ICoAC 2012, published in IEEE Xplore.
and complicated for finding a solution.
Various approaches can be used for service search, such as, searching in UDDI, Web and service portals, as discussed previously in the chapter 2. We take up the is- sue of web service searching and composition at service registriesfor its practicality in business world as providers would like to post their services centrally, as searching there is less time consuming than searching on world wide web.
Universal Description, Discovery and Integration (UDDI) (Carolgeyer (2013)) is an industry standard for service registries, developed to solve the web service search prob- lem. Basically, UDDI supports service search by name, location, business, bindings or tModels, and binds two services based on composability of their protocols (citeuddi).
Traditional UDDI based service search lacks the ability to recognize all the features described in WSDL files. Hence the search API(Application Program Interface) is lim- ited by the kind of information that is available and searchable in UDDI entries and do not provide any support for complex searches like input/output parameter based search and automatic composition of web services. Also, our experiments show that average response time of current UDDI implementations increase with a substantial increase in number of services registered.
The above shortcomings of UDDI motivated us tobuild an extended service reg- istry(ESR) system capable of offering powerful and efficient web service search and composition operations, by extending UDDI with meta information. We propose the use of Object Relational Database as repository of web services. Information about the web services, extracted from their WSDLs, are stored in tables and relational alge- braic operators are used for service search and composition.
A web service,ws, has typically two sets of parameters - set of inputswsI and set
of outputswsO. By service composition, we mean making of a new service(that does not exist on its own) from existing services registered in UDDI. Conventionally two serviceswsi andwsj are said to be composable iffwsOi =wsIj, i.e,wsjreceives all the required inputs from outputswsi has. A service composition is formed by constructing a chain of such composable services.
Recently, many researchers have utilized relational database techniques to solve the service composition problem (Kwonet al.(2007); Leeet al.(2011); Zenget al.(2010)) as discussed previously in chapter 2. These relational techniques can be applied to a large number of web services in that their scalability is not limited by size of physical memory. However the current approaches use multiple joins (Kwon et al. (2007);
Leeet al.(2011); Zenget al.(2010)) on tables to pre-compute all possible web service compositions, which may be unnecessary. Also most of the existing algorithms (Kwon et al.(2007); Lee et al.(2011); Zenget al.(2010)) construct chains of services based onexact matches of input/output parameters to satisfy a given query, as explained earlier. However, this approach fails when the available services satisfy only a part of the input/output parameters in the given query.
When a user is looking for a web service for given input and desired output param- eters and there is no single web service in the service registry satisfying the request, service Composition becomes necessary. However, the making of a service composi- tion chain may fail at a point when the output parameters of a preceding service (wsOP) does not match exactly with the input parameters of a succeeding service(wsIS). We propose an approach to alleviate this problem by making match criteria flexible.
In addition to Exact match we allow Partial as well as Super match for conditions wsOi ⊂wsIj andwsOi ⊃wsIj respectively(Lakshmi and Mohanty (2012)). Further, to avoid the need for multiple joins and to speed up querying,we propose to use multi- valued attributes for storing input and output parameters in our database design, and hence use an Object Relational database for our proposedESR. Based on this con-
cept we haveexplored the feasibility of input/output based web service Search and Composition in our proposed ESR system, to support varying requirements of the consumer.
In this chapter we first describe the types of service matches, extended from the clas- sical definition of service match in section3.2. In section 3.3 that follows, we propose an extended service registryusing ORDBMS approach. The database schema is explained in detail and an argument on the need for ORDBMS is also given in this section. Fol- lowing is section 3.4 that provides a discussion on the performance and scalability of our extended service registry in comparison with the standard UDDI service registry.
Conclusion is given in section 3.5.