Erasure Coding

MinIO Erasure Coding is a data redundancy and availability feature that allows MinIO deployments to automatically reconstruct objects on-the-fly despite the loss of multiple drives or nodes in the cluster. Erasure Coding provides object-level healing with significantly less overhead than adjacent technologies such as RAID or replication.

MinIO partitions each new object into data and parity shards, where parity shards support reconstruction of missing or corrupted data shards. MinIO writes these shards to a single erasure set in the deployment. MinIO can use either data or parity shards to reconstruct an object, as long as the erasure set has read quorum. For example, MinIO can use parity shards local to the node receiving a request instead of specifically filtering only those nodes or drives containing data shards.

Since erasure set drives are striped across the server pool, a given node contains only a portion of data or parity shards for each object. MinIO can therefore tolerate the loss of multiple drives or nodes in the deployment depending on the configured parity and deployment topology.

At maximum parity, MinIO can tolerate the loss of up to half the drives per erasure set (\((N / 2) - 1\)) and still perform read and write operations. MinIO defaults to 4 parity shards per object with tolerance for the loss of 4 drives per erasure set. For more complete information on selecting erasure code parity, see Erasure Code Parity (EC:N).

Use the MinIO Erasure Code Calculator when planning and designing your MinIO deployment to explore the effect of erasure code settings on your intended topology.

Erasure Sets

An Erasure Set is a group of drives onto which MinIO writes erasure coded objects. MinIO randomly and uniformly distributes the data and parity shards of a given object across the erasure set drives, where a given drive has no more than one block of either type per object (no overlap).

MinIO automatically calculates the number and size of Erasure Sets (“stripe size”) based on the total number of nodes and drives in the Server Pool, where the minimum stripe size is 2 and the maximum stripe size is 16. All erasure sets in a given pool have the same stripe size, and MinIO never modifies nor allows modification of stripe size after initial configuration. The algorithm for selecting stripe size takes into account the total number of nodes in the deployment, such that the selected stripe allows for uniform distribution of erasure set drives across all nodes in the pool.

Erasure set stripe size dictates the maximum possible parity of the deployment. Use the MinIO Erasure Coding Calculator to explore the possible erasure set size and distributions for your planned topology. MinIO strongly recommends architecture reviews via MinIO SUBNET as part of your provisioning and deployment process to ensure long term success and stability. As a general guide, plan your topologies to have an even number of nodes and drives where both the nodes and drives have a common denominator of 16.

Erasure Code Parity (`EC:N`)

MinIO uses a Reed-Solomon algorithm to split objects into data and parity shards based on the Erasure Set size in the deployment. For a given erasure set of size M, MinIO splits objects into N parity shards and M-N data shards.

MinIO uses the EC:N notation to refer to the number of parity shards (N) in the deployment. MinIO defaults to EC:4 or 4 parity shards per object. MinIO uses the same EC:N value for all erasure sets and server pools in the deployment.

MinIO can tolerate the loss of up to N drives per erasure set and continue performing read and write operations (“quorum”). If N is equal to exactly 1/2 the drives in the erasure set, MinIO write quorum requires \(N + 1\) drives to avoid data inconsistency (“split-brain”).

Setting the parity for a deployment is a balance between availability and total usable storage. Higher parity values increase resiliency to drive or node failure at the cost of usable storage, while lower parity provides maximum storage with reduced tolerance for drive/node failures. Use the MinIO Erasure Code Calculator to explore the effect of parity on your planned cluster deployment.

The following table lists the outcome of varying erasure code parity levels on a MinIO deployment consisting of 1 node and 16 1TB drives:

Outcome of Parity Settings on a 16 Drive MinIO Cluster
Parity	Total Storage	Storage Ratio	Minimum Drives for Read Operations	Minimum Drives for Write Operations
`EC: 4` (Default)	12 Tebibytes	0.750	12	12
`EC: 6`	10 Tebibytes	0.625	10	10
`EC: 8`	8 Tebibytes	0.500	8	9

Storage Classes

MinIO supports redundancy storage classes with Erasure Coding to allow applications to specify per-object parity. Each storage class specifies a EC:N parity setting to apply to objects created with that class.

MinIO storage classes for erasure coding are distinct from Amazon Web Services storage classes used for tiering. MinIO erasure coding storage classes define parity settings per object, while AWS storage classes define storage tiers per object.

Note

For transitioning objects between storage classes for tiering purposes in MinIO, refer to the documentation on lifecycle management.

MinIO provides the following two storage classes:

STANDARD

The STANDARD storage class is the default class for all objects. MinIO sets the STANDARD parity based on the number of volumes in the Erasure Set:

Erasure Set Size	Default Parity (EC:N)
5 or Fewer	EC:2
6 - 7	EC:3
8 or more	EC:4

You can override the default STANDARD parity using either:

The MINIO_STORAGE_CLASS_STANDARD environment variable, or
The mc admin config command to modify the storage_class.standard configuration setting.

The maximum value is half of the total drives in the Erasure Set. The minimum value is 2.

STANDARD parity must be greater than or equal to REDUCED_REDUNDANCY. If REDUCED_REDUNDANCY is unset, STANDARD parity must be greater than 2.

REDUCED_REDUNDANCY

The REDUCED_REDUNDANCY storage class allows creating objects with lower parity than STANDARD. REDUCED_REDUNDANCY requires at least 5 drives in the MinIO deployment.

MinIO sets the REDUCED_REDUNDANCY parity to EC:2 by default. You can override REDUCED_REDUNDANCY storage class parity using either:

The MINIO_STORAGE_CLASS_RRS environment variable, or
The mc admin config command to modify the storage_class.rrs configuration setting.

REDUCED_REDUNDANCY parity must be less than or equal to STANDARD.

MinIO references the x-amz-storage-class header in request metadata for determining which storage class to assign an object. The specific syntax or method for setting headers depends on your preferred method for interfacing with the MinIO server.

For the mc command line tool, certain commands include a specific option for setting the storage class. For example, the mc cp command has the storage-class option for specifying the storage class to assign to the object being copied.
For MinIO SDKs, the S3Client object has specific methods for setting request headers. For example, the minio-go SDK S3Client.PutObject method takes a PutObjectOptions data structure as a parameter. The PutObjectOptions data structure includes the StorageClass option for specifying the storage class to assign to the object being created.

Bit Rot Protection

Silent data corruption or bit rot is a serious problem faced by data drives resulting in data getting corrupted without the user’s knowledge. The corruption of data occurs when the electrical charge on a portion of the drive disperses or changes with no notification to or input from the user. Many events can lead to such a silent corruption of stored data. For example, ageing drives, current spikes, bugs in drive firmware, phantom writes, misdirected reads/writes, driver errors, accidental overwrites, or a random cosmic ray can each lead to a bit change. Whatever the cause, the result is the same - compromised data.

MinIO’s optimized implementation of the HighwayHash algorithm ensures that it captures and heals corrupted objects on the fly. Integrity is ensured from end to end by computing a hash on READ and verifying it on WRITE from the application, across the network, and to the memory or drive. The implementation is designed for speed and can achieve hashing speeds over 10 GB/sec on a single core on Intel CPUs.