Toward digital twins' workload allocation on clouds with low-cost microservices streaming interaction

A. Tchernykh, A. Facio-Medina, B. Pulido-Gaytan, R. Rivera-Rodriguez, J.M. Cortes-Mendoza, G. Radchenko, M. Babenko, I. Chernykh, I. Kulikov, S. Nesmachnow

    Research output: Contribution to conference (No Proceedings)Paperpeer-review


    A Digital Twin (DT) is a set of computational models representing real-time physical objects and processes in a digital world. The increasing adoption of this paradigm by the major industrial equipment vendors to simulate real-time working conditions and perform smart decision-making, established the Smart Factory Digital Twins architectures, where a set of DTs published as microservices interact with each other exchanging information by streaming technology. In this sense, a fundamental problem consists of selecting adequate computational resources to simulate the physical objects. In an Infrastructure as a Service (IaaS) cloud for DTs, the allocation focuses on distributing the jobs of the set into the virtual machine instances in a way that computational resources demand is satisfied and the cost is minimized. In this paper, we propose a set of algorithms based on heuristics, metaheuristics, and Mixed Integer Programming to find low-cost solutions. The performance of algorithms is evaluated using Amazon EC2 instances and DT jobs with randomly generated bandwidth, memory, and processor requirements. The experimental results show that the proposed approaches based on bin packing, genetic algorithms, partition, filtering, set coverage, and branch and bound strategies present a competitive performance in the workload allocation of the computational set of jobs of a DT into an IaaS cloud environment. Our allocation heuristic-based techniques allow considerable cost savings in medium and large periods concerning standard approaches such as local search. © 2020 IEEE.
    Original languageEnglish
    Number of pages7
    Publication statusPublished - 2020
    Event2020 Ivannikov Ispras Open Conference - Moscow, Russian Federation
    Duration: 10 Dec 202011 Dec 2020


    Conference2020 Ivannikov Ispras Open Conference
    Abbreviated titleISPRAS 2020
    Country/TerritoryRussian Federation


    • Digital Twin
    • EC2
    • Evolutionary Algorithms
    • Instance Provisioning
    • Metaheuristics
    • Mixed Integer Programming
    • Smart Factory
    • Costs
    • Decision making
    • Digital twin
    • Genetic algorithms
    • Integer programming
    • Multimedia systems
    • Competitive performance
    • Computational model
    • Computational resources
    • Industrial equipment
    • Mixed integer programming
    • Performance of algorithm
    • Streaming technology
    • Workload allocation
    • Infrastructure as a service (IaaS)


    Dive into the research topics of 'Toward digital twins' workload allocation on clouds with low-cost microservices streaming interaction'. Together they form a unique fingerprint.

    Cite this