#Core Concepts

• Erasure Coding (EC): MinIO employs Reed-Solomon erasure coding to break objects into data and parity shards, distributing them across multiple drives to ensure fault tolerance. For example, in a 16-drive setup, data can be split into 12 data shards and 4 parity shards, allowing the system to rebuild data even if up to 4 drives fail.

• Server Pools & Erasure Sets: MinIO Server Pools are logical groupings of storage resources, where each pool consists of multiple nodes sharing storage and compute capabilities. Within a pool, drives are automatically organized into one or more Erasure Sets.

  • Data Distribution: When an object is stored, it is split into data and parity shards and distributed across different drives within an erasure set.
  • Redundancy Model: Erasure sets form the fundamental unit of data redundancy, ensuring resiliency based on configured data-to-parity shard ratios.
  • Scalability: A single MinIO storage pool can contain multiple erasure sets, and new data is always written to the erasure set with the most available capacity.