Testbeds and Research Infrastructures for the Development of Networks and Communities. 11th International Conference, TRIDENTCOM 2016, Hangzhou, China, June 14-15, 2016, Revised Selected Papers

Research Article

Towards an Experimental LegoLand: Slice Modification and Recovery in ExoGENI Testbed

Download48 downloads
  • @INPROCEEDINGS{10.1007/978-3-319-49580-4_4,
        author={Yufeng Xin and Ilya Baldin and Anirban Mandal and Paul Ruth and Jeff Chase},
        title={Towards an Experimental LegoLand: Slice Modification and Recovery in ExoGENI Testbed},
        proceedings={Testbeds and Research Infrastructures for the Development of Networks and Communities. 11th International Conference, TRIDENTCOM 2016, Hangzhou, China, June 14-15, 2016, Revised Selected Papers},
        proceedings_a={TRIDENTCOM},
        year={2017},
        month={1},
        keywords={},
        doi={10.1007/978-3-319-49580-4_4}
    }
    
  • Yufeng Xin
    Ilya Baldin
    Anirban Mandal
    Paul Ruth
    Jeff Chase
    Year: 2017
    Towards an Experimental LegoLand: Slice Modification and Recovery in ExoGENI Testbed
    TRIDENTCOM
    Springer
    DOI: 10.1007/978-3-319-49580-4_4
Yufeng Xin1,*, Ilya Baldin1, Anirban Mandal1, Paul Ruth1, Jeff Chase2
  • 1: University of North Carolina at Chapel Hill
  • 2: Duke University
*Contact email: yxin@renci.org

Abstract

This paper describes advanced capabilities that were deployed recently in the ExoGENI testbed to offer increased flexibility in provisioning, modifying, and recovering the topologies and the configuration settings of the virtual systems, or , in which experiments are run. Using the analogy of building complex structures with LEGO blocks, we envision an environment in which users arbitrarily scale out, scale in, scale up, and scale down their topologies using various modular constructs of compute, storage, and network resources. Portions of topologies can be shut down and brought back up to support resiliency, repeatability, migration, and other needs of the control software or application. Distributed applications running inside of slices can require programmatic control over the evolution of the topology as the execution progresses. The introduced capabilities, and , are used either with the user GUI or through the programmable APIs. These new features expand the range and ease of options available to cloud-control software and to application developers as they test their designs at scale.