Zoned storage devices are a class of storage devices with an address space that is divided into zones which have write constraints different from regular storage devices.
The zones of zoned storage devices must be written sequentially. Each zone of the device address space has a write pointer that keeps track of the position of the next write. Data in a zone cannot be directly overwritten. The zone must first be erased using a special command (zone reset). The figure below illustrates this principle.
Zoned Storage Devices Principle
Zoned storage devices can be implemented using various recording and media technologies. The most common form of zoned storage today uses the SCSI Zoned Block Commands (ZBC) and Zoned ATA Commands (ZAC) interfaces on Shingled Magnetic Recording (SMR) HDDs. ZBC and ZAC enable a zoned block storage model; SMR technology enables continued areal density growth to meet the demands for expanding data needs, and requires the zoned block access model.
Solid State Disks (SSD) storage devices can also implement a zoned interface to reduce write amplification, reduce the device DRAM needs and improve quality of service at scale. The NVMe Zoned NameSpace (ZNS) is a technical proposal of the NVMe standard committee aiming at adding a zoned storage interface to the NVMe interface standard. NVMe ZNS is currently in the process of being standardized.
Linux Ecosystem Support¶
Zoned storage devices are not plug-and-play replacement of traditional storage devices due to the sequential write constraints of zones. Special software and drivers are required to ensure compliance of the application operation to the device constraints.
Support for zoned storage device was added to the Linux® kernel with the release of version 4.10.0 and enables support for zoned storage at different levels (disk driver, file system, device mapper drivers), offering a wide range of options for supporting applications. This support is based on the Zoned Block Device (ZBD) abstraction.
A zoned block device is a generic representation of a zoned storage device independent of the device access protocol and interface. This abstraction is the basis of Linux kernel support for zoned storage.
The interface associated with the ZBD device abstraction is an extension to the traditional Linux block device interface. The ZBD interface, combined with device drivers, provide to kernel subsystems (e.g. File systems) and to user applications a generic zone management interface compatible across all zoned device types and access protocols.
A simplified view of the kernel structure including the ZBD interface is shown in the figure below.
Linux kernel Zoned Storage Device Support Overview
Linux ZBD interface implementation provides functions to discover the zone configuration of a zoned device and functions to manage zones (e.g. Zone reset). Furthermore, the Linux kernel ZBD support also modifies the kernel block I/O stack to ensure that the device access constraints (zone spanning commands, sequential write ordering, etc) are met.
Developing for Zoned Storage¶
There are several ways to design a system application for zoned storage depending on one's system structure and ability to modify the application layer and Linux kernel being used.
Users that own software applications and do not depend on the operating system and file system to control the device can directly use the device interface protocol to issue zone management commands using a passthrough interface. In such case, the applications will need to be re-written with the new command sets as well as ensuring that all data streams are sequential. The libzbc library provides functions facilitating the implementation of applications using such approach.
Managing zoned storage directly from the application layer is a valid approach but can be difficult to implement, particularly for use cases where several device features must be combined (e.g. Using I/O priorities). Application level support can be simplified by relying on the kernel ZBD support which allows accessing and managing zoned devices using regular POSIX system calls. While this does not remove the zoned device access constraints from the application scope, implementation of sequential write streams and zone management can be simplified.
If a user lacks the software control at the application level while retaining control over Linux kernel version choices, more advanced kernel support features such as ZBD compliant file systems can be used to hide zoned storage access constraints from the application. Users may also rely on a device mapper driver exposing zoned storage devices as regular block devices. With such solution, existing file systems can be used without any modification.
More information on the features provided by the Linux kernel for different versions can be found here.
To get started with zoned storage, the Getting Started with SMR section details the first step necessary to setup and verify a system using SMR disks. The Linux Distributions section provides information regarding the availability of the ZBD interface on various Linux distributions. Various tools and open source projects supporting zoned storage are documented in the Applications and Libraries section.