Friday, January 26, 2018

Features of Hadoop

Parallel Execution
  • Extremely good at high-volume batch processing because of the ability to do parallel processing.
  • Can perform 10 times faster than on a single thread server or on mainframe
Data Locality
  • Data is not moved .
  • Processing data where it resides
      Note: This is the ideal choice, however it might not be possible to always achieve data locality

Fault Tolerance
  • Data is stored in HDFS, where data automatically gets replicated into 2 other locations. 
  • Level of replication is configurable and this makes it incredibly reliable data storage system
  • Generates cost benefits by bringing massively parallel computing to commodity servers
  • A rough cost of hadoop inluding hardware, software and other expenses come to about 1,000 $ a terabyte about 1/5th to 1/20th of the data management system.
  • Open source platform and runs on industry-standard platform
  • New nodes can be easily added as data volume of processing needs grow without altering anything in the existing system and program

No comments:

Post a Comment