Hadoop – Modes of accessing HDFS

Modes of accessing HDFS: You can access Hadoop Distributed File System in 2 different ways.

1. Direct Access
2. Proxy based Access

Direct Access can be through ..

1. Interactive Unix Shell commands
2. Through Java and C++ APIs, clients retrieve metadata such as blocks locations from Namenode. Client directly access datanodes. HDFS accessing  through API is used in Map Reduce jobs.

Proxy Based Access through…

Clients communicated through a Proxy and it is strives to be language independent.

Several Proxy servers are packaged with Hadoop are

1. Thrift – Interface definition language.
2. WebHDFS REST – You can access HDFS through web, and the response is formatted in JSON, XML or Protocol Buffers.
3. Avro – Data serialization mechanism.