Zarr A Cloud-Optimized Storage for Interactive Access of Large Arrays

Abstract

For decades, the sharing of large N-dimensional datasets has posed issues across multiple domains. Interactively accessing terabyte-scale data has previously required significant server resources to properly prepare cropped or down-sampled representations on the fly. Now, a cloud-native chunked format easing this burden has been adopted in the bioimaging domain for standardization. The format — Zarr — is potentially of interest for other consortia and sections of NFDI.