Automating Metadata Web Service Deployment for Problem Solving Environments Ozgur Balsoy, Ying Jin, Galip Aydin, Marlon Pierce, and Geoffrey Fox Computer Science Department, Florida State University, Tallahassee, FL 32306 [email protected] Community Grids Lab, Indiana University, 501 N. Morton Street, Suite 224, Bloomington, IN 47404-3730 {Yinjin, Gaydin, Marpierc, Gcf}@indiana.edu Abstract. XML-based metadata information services are a crucial core service needed by Problem Solving Environments built over emerging service-based, globally-scaled distributed systems, as envisioned by the Open Grid Services Architecture and the Semantic Web. Developing user interfaces and services bindings for manipulating instances of particular schemas is thus extremely important and needs to be made as simple as possible. In this paper we describe procedures for automating the creation of Web Service environments that can be used to simplify the creation and deployment of schema-based metadata services. 1 Introduction One of the most important underpinnings of Problem Solving Environments (PSEs) is the management of diverse types of information. Grid computing and distributed Web services are metadata rich: all entities should be described by XML metadata. Metadata must be used to describe the various parts that must be managed by the PSE, such as available applications, service interfaces and their binding points, batch queuing systems, hardware profiles, and so on. Many projects have attempted to define metadata pieces needed to build service grids, and such pieces will be needed as grid service data in the Open Grid Service Architecture [1,2]. In addition to the service metadata, there is also need to provide organizational metadata descriptors for the services themselves, and we have suggested data models for describing applications [3,4] and their associated service bindings. We thus anticipate that metadata richness will only increase. We may place these issues into larger contexts. Autonomic computing (see [5] for a description and [6,7] for sample academic projects) seeks to define the interaction of components in next generation computing at all scales. In the autonomic computing vision, components mean everything from the physical component parts of individual computers to the service components of globally-scaled service environments. The ideal autonomic system is self-aware, self-monitoring, adaptive, and self-healing, among other characteristics.2 Ozgur Balsoy, Ying Jin, Galip Aydin, Marlon Pierce, and Geoffrey Fox Developing standards for Web services for computational grids is a major (and somewhat independent) component of the autonomic computing vision. Web services are somewhat loosely defined, but may be characterized in general as using existing Web technologies and standards (such as HTTP and XML) to build the distributed computing environments. Important Web service standards include the Web Service Description Language (WSDL) [8] and the Simple Object Access Protocol (SOAP) [9]. The Open Grid Service Architecture, developed under the auspices of the Global Grid Forum and primarily led by IBM and the Globus team, seeks to extend the capabilities of Web services in order to support more sophisticated services. The key to this development is the extension of WSDL portTypes to include metadata and service state information. In a major alternative effort, other groups are using various XML standards of the World Wide Web Consortium define what it terms the Semantic Web [10]. The Semantic Web vision borrows ideas from federated information systems and artificial intelligent, modified and simplified to apply to the much more ambitious problem of intelligent information retrieval and service execution on a global scale. The cornerstones of the Semantic Web are metadata descriptions (RDF) [11] and ontologies that provide semantic meaning. XML-based information services are the key to both the OGSA and the Semantic Grid. We may go a step further and claim that the there is a general need for many informal, lightweight information management systems for particular schemas. We have developed such a system, described in Refs [12] and [13], which provides several per-schema implementations: newsgroups, citation and reference managers, training and conference registration systems. All such systems possess the same general capabilities: browsing, search, “push” and “pull” style event delivery options, and access controls on information and publishing. More sophisticated systems that support schema-spanning searches (with the associated meta-ontology research problems [14,15]) may be built on top of this practical base. Given the demonstrable importance and proliferation of XML metadata, it becomes necessary to automate the creation of publication and editing tools for particular schemas. We describe the details of two such systems, which we refer to as “Schema Wizards”. The first is designed to simplify the creation of XML wizards for specific schemas. These wizards are in turn used for creating and publishing instances in the XML messaging system described in [13]. The Schema Wizard is thus a wizard for wizards. The second wizard system, a follow-on to the first, is designed to automate metadata-based Web services (such as service data associated with Grid Web services). In both cases, a particular schema is used as the underlying data model, and the wizard generates the user interface (view) and action bindings (control) necessary to create and deploy the new metadata-based service. 2 Schema Wizard: Automating Metadata Creation for Publication Applications that guide users through a complicated task are usually called wizards. XML wizards are applications that receive user inputs to generate XML documents. In our research on XML information frameworks (see particularly Ref [13]), XML isAutomating Metadata Web Service Deployment for Problem Solving Environments 3 used as the medium for storing information, but the systems are user driven. Wizards help users generate schema-based XML content and publish into an XML messaging broker. Such documents are then typically published to various sources, including both persistent repositories and listeners demanding immediate notification. In our current implementation, our information management system provides general purpose tools that can be used to create specific applications that are tied