drives.  Therefore, the number of drives you present must be a multiple of one of these numbers.  Each object is written to a single erasure-coding set.

Minio uses the largest possible EC set size which divides into the number of drives given. For example, *18 drives* are configured as *2 sets of 9 drives*, and *24 drives* are configured as *2 sets of 12 drives*.  This is true for scenarios when running MinIO as a standalone erasure coded deployment. In [distributed setup however node (affinity)...

two parity drives, 100 MiB file takes only approximately 114 MiB.

Below is a list of data/parity drives and corresponding _approximate_ storage space usage on a 16 drive MinIO deployment. The field _storage
usage ratio_ is simply the drive space used by the file after erasure-encoding, divided by actual file size.

| Total Drives (N) | Data Drives (D) | Parity Drives (P) | Storage Usage Ratio |

import "context"

const (
	healthDrivesOfflineCount = "drives_offline_count"
	healthDrivesOnlineCount  = "drives_online_count"
	healthDrivesCount        = "drives_count"
)

var (
	healthDrivesOfflineCountMD = NewGaugeMD(healthDrivesOfflineCount,
		"Count of offline drives in the cluster")
	healthDrivesOnlineCountMD = NewGaugeMD(healthDrivesOnlineCount,
		"Count of online drives in the cluster")

}

func (m *MetricValues) setDriveBasicMetrics(drive madmin.Disk, labels []string) {
	m.Set(driveUsedBytes, float64(drive.UsedSpace), labels...)
	m.Set(driveFreeBytes, float64(drive.AvailableSpace), labels...)
	m.Set(driveTotalBytes, float64(drive.TotalSpace), labels...)
	m.Set(driveUsedInodes, float64(drive.UsedInodes), labels...)
	m.Set(driveFreeInodes, float64(drive.FreeInodes), labels...)

### 1.2 Host Multiple Tenants on Multiple Drives (Erasure Code)

Use the following commands to host 3 tenants on multiple drives:

```sh
minio server --address :9001 /disk{1...4}/data/tenant1
minio server --address :9002 /disk{1...4}/data/tenant2
minio server --address :9003 /disk{1...4}/data/tenant3
```

			buckets = append(buckets, BucketInfo{
				Name:    v.Name,
				Deleted: v.Created,
			})
		}
	}

	return buckets, nil
}

func cloneDrives(drives []StorageAPI) []StorageAPI {
	newDrives := make([]StorageAPI, len(drives))
	copy(newDrives, drives)
	return newDrives
}

func getBucketInfoLocal(ctx context.Context, bucket string, opts BucketOptions) (BucketInfo, error) {
	globalLocalDrivesMu.RLock()

		mountPointCache.Store(path, false)
		return false
	}

	switch windows.GetDriveType(&wvolume[0]) {
	case windows.DRIVE_FIXED, windows.DRIVE_REMOVABLE, windows.DRIVE_REMOTE, windows.DRIVE_RAMDISK:
		// Recognize "fixed", "removable", "remote" and "ramdisk" drives as proper drives
		// which can be treated as an actual mount-point, rest can be ignored.

| Minimum number of drives per server when server count is 1      | 02            |
| Minimum number of drives per server when server count is 2 or 3 | 01            |
| Minimum number of drives per server when server count is 4      | 01            |
| Maximum number of drives per server                             | no-limit      |
| Read quorum                                                     | N/2           |

If you have traditional spinning (hdd) drives, some applications with high concurrency might require MinIO cluster to be tuned such that to avoid random I/O on the drives. The way to convert high concurrent I/O into a sequential I/O is by reducing the number of concurrent operations allowed per cluster. This allows MinIO cluster to be operationally resilient to such workloads, while also making sure the drives are at optimal efficiency and responsive.

If one or more drives are offline at the start of a PutObject or NewMultipartUpload operation the object will have additional data
protection bits added automatically to provide the regular safety for these objects up to 50% of the number of drives.
This will allow normal write operations to take place on systems that exceed the write tolerance.