Automated discovery of information services in heterogeneous distributed networks

Abstract

The Global Information Grid (GIG) will be comprised of collections of different service capability domains (SCDs). Each SCD offers a set of information services, such as voice over IP (VoIP), video delivery, and information translation, and is managed as a separate system. The GIG information services will include several types of communications services (e.g., VoIP and streaming video), translation services (e.g., document translation and data translation) and information services (e.g., content discovery and domain name service). These different services may be described using various methods, including specification documents and lookup tables, and will have associated service level agreements (SLAs). As SCDs become richer in their service offerings and more dynamic in their service availability, the discovery of end-to-end services meeting end-user needs becomes extremely challenging. Currently manual methods are employed to map end-user needs to the end-to-end service combinations that GIG deployments can support. Such manual methods are inefficient and error prone. Automation of end-to-end service discovery within the GIG is highly desirable, but is an exceedingly complex task. Current efforts to automate service discovery, for example, the service location protocol or the service oriented architecture, provide service discovery through registration and strict service type definitions. These require coordination across all SCDs for all possible information technology (IT) service offerings. Ideally, individual SCDs would describe their services through individual service description documents. Then mapping of end-user service requests to appropriate collections of SCD services could be performed automatically. This requires the development of semantic reasoning and ontology for service descriptions and service capability matching. This paper describes our approach for automated discovery of information services on GIG-like network deployments.