So, I've just worked my way through the GFS whitepaper (mapreduce is next), and was wondering what the major design/architecture differences are compared with Hadoop's DFS.
The JavaDoc points out one: single-writer, stream only (no record append, no writing to specific spot in file, etc). Is that a different design decision, or simply something that hasn't been implemented yet? Have you guys checked out the BigTable talk? http://norfolk.cs.washington.edu/htbin-post/unrestricted/colloq/details.cgi?id=437 -Yonik
