A parallel file system spreads files across many storage servers and disks so that numerous clients can read and write different parts simultaneously, delivering aggregate bandwidth far beyond a single server. It presents one unified namespace while striping data behind the scenes. In HPC and AI clusters it feeds thousands of GPUs during training and holds large checkpoints, making it a core high-performance storage tier.
Category: IT & Compute Infrastructure · Subcategory: Storage & Virtualization · Part of the AIDC Technical Terminology reference.