Skip to main content

How-To Guides

These guides cover installation, configuration and data verification for storage systems using host-managed SMR hard disks or NVMe Zoned Namespace (ZNS) SSDs.

Each zoned drive uses the XFS file system. XFS handles sequential zone writes, zone resets and garbage collection. The storage software uses ordinary file I/O and requires no zoned-device support.

Available Guides​

Storage SystemTypeWhat Goes on Zoned Storage
MinIOS3-compatible object storageObject data on every drive of the MinIO server pool
BeeGFSParallel cluster file systemFile contents on the storage targets
HDFSHadoop distributed file systemDataNode block files
JuiceFSPOSIX file system on object storageFile data chunks in the object store
SeaweedFSDistributed object and file storageVolume server volume files
GarageS3-compatible object storageObject data blocks, one data directory per drive
RustFSS3-compatible object storageObject data, object metadata and internal state, one file system per drive
AIStoreObject storageObject files on the target mountpaths
GlusterFSDistributed file systemBrick file contents
LonghornBlock storage for KubernetesVolume replica files (V1 data engine)
CubeFSDistributed file system (POSIX and S3)DataNode extent files (replica volumes)
Apache OzoneDistributed object store (S3, Hadoop)Datanode container block files

Separate metadata services, such as HDFS NameNodes, Ozone Managers, JuiceFS metadata engines and the Garage metadata database, use conventional storage, typically SSDs. MinIO and RustFS store their object metadata alongside the data on the zoned drives.

Before You Start​

System Requirements
  • Linux kernel: 6.15+ with CONFIG_XFS_FS and CONFIG_XFS_RT enabled
  • xfsprogs: 6.15+

See Linux Distributions and Getting Started for kernel and package requirements.

Without zoned hardware, use an emulated device in place of a physical drive. See Zoned Block Device Emulation for zloop and null_blk setup.