Reputation: 109
I am a young researcher and plan to buy a computing server for (maybe "big") data analysis. The server will have a 20TB hard disk space. My current question is should I use normal linux file system or hadoop hdfs as the system. Would anybody kindly explain the pros and cons on these two choices?
Upvotes: 0
Views: 139
Reputation: 294297
It really makes no sense to have HDFS over a single node (one server). HDFS is all about distributing data so that compute tasks run close to the data, and having redundant copies of the data to be able to tolerate hardware failures. A single node will offer not a single HDFS benefit. If you have a cluster of machines (eg. 10 servers) then yes, you can ask this question. As is, HDFS is not an option.
Upvotes: 1