Virtual Infrastructure Twin for Computing-Instrument Ecosystems: Software and Measurements

TL;DR: In this article , the authors present a Virtual Infrastructure Twin (VIT) of an ecosystem to support the software development and testing for multi-site computing-instrument ecosystems, implemented using virtual hosts connected over emulated local-area networks at sites, which are in turn connected over an emulated wide-area network.

Abstract: Advanced science workflows are orchestrated over complex ecosystems consisting of computing platforms and physical instruments connected over wide-area networks. The availability of physical facilities of these ecosystems may be limited during typically long software development periods, due to the expense, disruption risk and limited readiness during the design and construction stages. We present a Virtual Infrastructure Twin (VIT) of such an ecosystem to support the software development and testing for multi-site computing-instrument ecosystems. A VIT is implemented using virtual hosts connected over emulated local-area networks at sites, which are in turn connected over an emulated wide-area network. We describe the design and implementation details of a VIT including: (i) host and network emulations and the incorporation of instrument simulators, (ii) services for remote instrument steering, and (iii) performance analytics based on network throughput and disk input/output measurements. We illustrate a use case by emulating an ecosystem of four geographically distributed laboratory sites with services for accessing and steering instruments and estimating throughput profiles using physical testbed measurements.