DepTwin: A Digital Twin-Based Framework to Manage Deployments in Cloud

Abstract

In the ever-evolving context of the digital twin and Industry 4.0 landscape, it has become increasingly evident that there is a need for a comprehensive framework to maintain the deployed systems that make those cyber–physical systems vital. Deployment management in the cloud-based continuum has become an architecturally complex endeavor to make observable, fault-tolerant, and regulated. Although elaborate Standard Operating Procedures (SOPs) and sophisticated methodologies define how deployments should be managed in a cloud-based environment, none of them has conclusively determined a framework that ensures minimal human intervention; this is where DepTwin attempts to create a digital replica of systems to handle faults, regulate orchestration and enhance observability. The proposed framework, DepTwin, defines a clear digital twin-based methodology to ensure management of deployed applications in cloud-based environments. It leverages both empirical methods in the form of the proposed Optimal Cloud Resource Identification Algorithm (OCRI) along with ever-evolving large language model-based agents to enhance observability, handle faults, and support resource provider identification in multi-cloud deployment scenarios. DepTwin not only improves application availability and reduces operational overhead but also improves overall resilience to meet Service Level Agreements (SLAs). Evaluated on the iTrust2 health record system across 213 trials with 22 code faults and three resource-degradation scenarios, DepTwin localized faults in 92.4% of trials, repaired 66.7%, and restored SLA compliance in every resolvable resource scenario.