Every VFX and animation shot lives on shared storage — but when render farms burst into the cloud, that storage suddenly has to feed thousands of nodes across a long-haul link back to the studio. KNFSD File Cache is an open source NFS caching solution built for exactly this problem: it re-exports a studio's on-prem NFS source through the Linux kernel and caches actively-read assets on local NVMe in AWS, so texture sets, cached simulations, and reference plates cross the WAN once and then serve an entire cloud render fleet at local speed — over standard NFS, with no proprietary drivers or client-side agents to roll out across render nodes.
Sean Wallitsch walks through forking and hardening the open-source knfsd-cache-utils project for production VFX/animation workloads: modernizing the codebase, remediating a steady stream of CVEs, performance-tuning the Linux kernel, network, and NFS stack for the read-heavy, bursty I/O patterns typical of frame rendering, adding developer-focused tooling, and expanding OpenTelemetry observability so studio infrastructure teams can see exactly what their render farm is waiting on.
