TR-13-02 Using Semantic Web Description Techniques for Managing Resources in a Multi-Domain Infrastructure-as-a-Service Environment

Yufeng Xin (RENCI), Ilia Baldine (RENCI), Jeff Chase (Duke University) and Kemafor Anyanwu (NCSU), Technical Report TR-13-02, Using Semantic Web Description Techniques for Managing Resources in a Multi-Domain Infrastructure-as-a-Service Environment, Renaissance Computing Institute, 2013.

This paper reports on experience with using Seman- tic Web technologies for managing multi-domain net- working infrastructure-s-a-service (IaaS) testbed. An OWL ontology based on newly-created vocabular- ies was used to model multi-layer network providers with common base classes for fundamental cyber- resources, and adaptation functions from resources at one layer onto resources of the same base class at the layer below. Extended SPARQL path queries supported by GLEEN were used to support topol- ogy embedding and resource provisioning for creat- ing connected arrangements of compute, storage and network resources gathered from multiple resource providers.
The context for the work is the use of the seman- tic models in ORCA – the control software for Ex- oGENI, a new testbed funded through NSFs GENI project. ExoGENI is a multi-domain cloud testbed with a high degree of control over networking func- tions, including links within each domain and dy- namic inter-domain links over national circuit fab- rics. The paper describes how the semantic network models enable ExoGENI to instantiate on-demand virtual topologies of virtual machines linked by on- demand circuits and segments for a variety of applica- tions ranging from networking experiments to high- performance computing.