作者: Adam Fried-Gintis , Dan Olster , Donald J. Molaro , David Fellinger
DOI:
关键词:
摘要: A failure resilient distributed replicated data storage system is described herein. The includes zones that are independent, and autonomous from each other. include nodes independent autonomous. devices. When a item stored, it partitioned into plurality of objects parity calculated. Reassembly instructions created for the item. spread across all in system. also zones. read request received, prepared lowest latency according to reassembly instructions. This provides resiliency while keeping amount space required relatively low.