- Availability -- Master process are single point of failure (Hadoop 2.x brings HA supports for NameNode and JobTracker to migrate this issue)
- Security -- security model is disabled by default. No storage or wire-level encryption.
- configure to run with Keberos (a network authentication protocol)
- HDFS -- lack of availability
- MapReduce -- is a "batch-based architecture" and it uses "shared-nothing architecture". Not good for job that needs real-time data access.
- Ecosystem version compatibilities
About Me
- Eddy Zhu
- Love JAVA related technologies. Recently researching on Enterprise Integration (SOA and Messaging), Mobility and Big Data. I have working in JAVA related technologies as Software Architect, Enterprise Architect and Software Developer/Engineer for over 11 years. Currently, I am working as Senior Consultant of VMWare Inc.
Friday, November 9, 2012
Hadoop limitations
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment