Breaking News

HDFS Introduction

A simple file system based on GFS (google file system)
ÄGFS was derived from big files paper authored by Larry Page and Sergey Brin in Stanford 
ÄPrinciple based on WORM (Write once and read many times)
ÄData copied from source and parallel processing performed on local dataset 
ÄHDFS setup needs commodity hardware –A big plus for cost optimization 
ÄHighly fault-tolerant 
ÄHigh throughput 
ÄSuitable for applications with large data sets 
ÄStreaming access to file system data 
ÄInterface to HDFS is patterned after Unix file system 

HDFS is not good for :
Ä    Low-latency data access,Lots of small files,Multiple writers, arbitrary file modifications 


- See more at:
Toggle Footer