Metaverse Distributed Storages: High Availability Quantification using Stochastic Reward Nets

The constantly evolving Metaverse, particularly advanced Augmented Reality (AR) and Virtual Reality (VR), demands a highly scalable and robust technological infrastructure to handle the substantial data generated by avatar interactions and immersive virtual environments. Centralised clouds are often inadequate to meet strict performance demands, highlighting the need for decentralised edge-based strategies. This paper proposes a Mobile Edge Computing (MEC)-oriented framework for distributed storage in Metaverse services, focusing on orchestrating heterogeneous Edge Nodes and efficient computational offloading. We present an innovative Stochastic Reward Net (SRN) model that thoroughly analyses the performance and resilience of this decentralised infrastructure. Our SRN model incorporates failure and recovery dynamics of Meta-Edge Storage Nodes, alongside High Availability (HA) and Live Migration (LM) protocols to protect microservices and minimise downtime across diverse edge components. Additionally, the model supports expansion alongside increasing Edge Nodes, affording flexible scalability for future Metaverse growth. Through rigorous sensitivity analysis and Discrete Event Simulation (DES), we assess crucial performance parameters, demonstrating how our MEC-based architecture and the SRN model jointly improve overall scalability, availability, and resource efficiency. These findings underline the capacity of a decentralised edge approach to sustain data-intensive Metaverse services, delivering a more fault-tolerant and resource-optimised alternative than conventional centralised solutions.