HDFS file system provides an API for an abstract File System Based on hadoop, which supports stream-based access to data in the file system.
Features:
1. Support for ultra-large files
2. Detect and quickly respond to hardware faults (fault detection and Automatic Recovery)
3. Streaming Data Access focuses on data throughput rather than data response speed
4. Simplified consistency model with one write and multiple reads.
Not Suitable:
5. Low-latency data access and real-time interaction with users. Because hadoop optimizes high data throughput and sacrifices data latency, hbase Or Cassandra can be considered for low-latency access.
1.1.1. HDFS Architecture
The number of normal horizontal replicas is the copy coefficient.
Namenode:
SNN:
Datanode:
Client:
Source code file: Under the org. Apache. hadoop. HDFS package
You can view the source code implementation when necessary.
2. Remotely called Interfaces
The main HDFS process can be analyzed in detail when necessary.