Cluster balancing algorithm
The HDFS Balancer runs in iterations. Each iteration contains the following four steps: storage group classification, storage group pairing, block move scheduling, and block move execution.
Scaling Namespaces and Optimizing Data Storage
Also available as:
- Introduction
- Scaling namespaces
- Optimizing data storage
- Balancing data across disks of a DataNode
- Increasing storage capacity with HDFS erasure coding
- Increasing storage capacity with HDFS compression
- Setting archival storage policies
- Balancing data across an HDFS cluster
- Optimizing performance
- Using the NFS Gateway for accessing HDFS
- Data storage metrics
- APIs for accessing HDFS