GFS Paper (2003)

Back to Key Papers to Read

“The Google File System” (2003) — Ghemawat, Gobioff, Leung

The distributed file system that started it all. Designed for large-scale data-intensive applications with commodity hardware. Key ideas: large chunk size (64MB), single master with multiple chunkservers, append-optimized writes. Directly inspired HDFS and the entire Hadoop ecosystem.


google-internal papers gfs #2003