AI models can be backed up to ensure their preservation and availability if there is data loss, system failures, or accidental changes. NVIDIA Triton Inference Server does not natively support backup and restore functionalities for model data and inference state. Triton Inference Server primarily focuses on efficiently serving AI models for inference and does not include integrated mechanisms for data backup and restoration. Models are typically stored in a repository in Persistent Volumes offered by Kubernetes. Customers can take advantage of backup solutions for Kubernetes to backup and restore AI models and inference data.
