Disk Balancer overview

Ozone disk balancer is a tool that evenly spreads data across different storage disks inside a single DataNode.

Apache Ozone effectively distributes containers across multiple DataNodes in a cluster, ensuring cluster-wide balancing. However, within a single DataNode, individual disks can still drift out of balance over their operational lifetime. Disk imbalance generally occurs due to:

• Adding new disks to expand a DataNode's storage capacity.
• Replacing broken disks with new, empty drives.
• Massive deletions of blocks or replicas that disproportionately empty certain disks.

Uneven disk utilization creates performance bottlenecks. Over-utilized disks become I/O hotspots, degrading the DataNode's overall throughput and increasing the risk of failure. To resolve this, Ozone features an intra-node Disk Balancer that automatically detects imbalances and evenly distributes data across all volumes within a single DataNode.