Speaker
Description
LHAASO generates massive volumes of scientific data, posing significant challenges for efficient data access, management, and distributed analysis. To address these challenges, we present JWanFS, a wide-area distributed object storage platform with a total capacity of 8.38 PB, spanning three geographically separated sites in Beijing, Huairou, and Daocheng. JWanFS provides a unified namespace for transparent cross-site data access, with latency-aware gateway selection to optimize data access paths. It supports multiple access methods, including web interfaces, command-line tools, cross-platform clients, POSIX-compatible file system mounts, and standard S3 APIs, offering a consistent view of scientific data across heterogeneous environments. Through S3-compatible SDKs for languages such as Python, Go, and Java, JWanFS enables seamless integration with existing scientific analysis workflows and supports parallel data access from distributed computing nodes. With scalable storage capacity, reliable data management, and flexible cross-site access, JWanFS aims to simplify scientific data handling and facilitate efficient, collaborative, and large-scale data processing within the LHAASO collaboration.