Skip to main content

Getting Started with SMR Hard Disks

Hard disk drives that use Shingled Magnetic Recording technology can have different interface implementations. This results in different usage models:

  • Drive Managed Interface: SMR disks that implement this interface are seen by the host kernel and by applications as identical to regular disks. Most Linux® kernels recognize these disks. SMR disks that implement this interface make no LBA-space zoning information available to the host. Drive Managed disks are therefore not considered to be zoned block devices.

  • Zoned Block Interface: SMR hard-disk drives that implement the ZBC and ZAC feature sets also provide commands to the host that allow the host to identify and control the device's zones. This interface has two different variations, or model implementations (See SMR Interface Implementations). Those two model implementations are:

    • Host Aware: This zone model offers the convenience and flexibility of Drive Managed disks (for example, it provides random write capabilities) and also supports the full set of zone commands defined in the ZBC and the ZAC standards. Host Aware disks can support both the regular block device abstraction ("regular disk") and the zoned block device abstraction.

    • Host Managed: This zone model defines a device type that is different from the "regular disk" device type. Host Managed disks can be used only as zoned block devices. This is necessary in order to satisfy the strong sequential write constraints defined by this zone model.

In the following sections, Host Aware disk models are considered to have characteristics similar to Host Managed drives. Sequential writes are therefore assumed to be a constraint for the correct operation of drives that implement the Host Aware disk model.

Serial ATA ZAC Disks and SATA Host Controllers​

Serial ATA (SATA) host adapters, including those that use the Advance Host Controller Interface (AHCI) standard, are able to scan and initialize connections with Host Aware disks. Most AHCI host adapters are known to work with Host Managed disk drives (this is because the adapter usually does not react to the device signature of the connected disk).

Verifying The Disk​

On systems that have zoned block device support enabled, SATA host aware disks and host managed disks can be connected directly to SATA ports on the host controller.

After booting the system, use the lsscsi utility to list all SCSI devices attached to the system and verify the presence of the newly-connected disk:

# lsscsi -g
[2:0:0:0]    disk    ATA      INTEL SSDSC2CT18 335u  /dev/sda   /dev/sg0
[5:0:0:0]    zbc     ATA      HGST HSH721415AL T220  /dev/sdb   /dev/sg1

In this example, the disk '/dev/sda' is the system boot disk and the disk '/dev/sdb' is recognized as a ZBC disk.

The second column of lsscsi output indicates the device type. The value zbc is always used for Host Managed ZBC and ZAC disks. This corresponds to the ZBC-defined device type 0x14 for SAS disks and corresponds to the ZAC-defined device signature 0xabcd for SATA disks. Because Host Aware disks have the same device type or device signature as regular disks, lsscsi lists host aware disks as disk.

note

The lsscsi utility does not list SATA ZAC disks with the type zac. The type zbc is always used, because the kernel internally implements a SCSI-to-ATA translation layer (SAT), which allows SATA devices to be represented as SCSI devices.

Checking The Disk Information​

Verify the zone model of the disk by checking the zoned sysfs attribute:

# cat /sys/block/sdb/queue/zoned
host-managed

The possible values of the zoned attribute are shown in the table below.

ValueDescription
noneRegular disk or drive managed ZBC/ZAC disk
host-awareHost aware ZBC/ZAC disk
host-managedHost managed ZBC/ZAC disk

Kernel messages also contain useful information about the disk:

# dmesg
ahci 0000:00:11.5: version 3.0
ahci 0000:00:11.5: AHCI 0001.0301 32 slots 2 ports 6 Gbps 0x3 impl SATA mode
ahci 0000:00:11.5: flags: 64bit ncq sntf led clo only pio slum part ems deso sadm sds apst
...
scsi host5: ahci
ata5: SATA max UDMA/133 abar m524288@0x9d100000 port 0x9d100200 irq 55
ata5: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
ata5.00: ATA-9: HGST HSH721414ALN6M0, L4GMT220, max UDMA/133
ata5.00: 27344764928 sectors, multi 0: LBA48 NCQ (depth 32), AA
ata5.00: configured for UDMA/133
...
sd 5:0:0:0: Attached scsi generic sg1 type 20
sd 5:0:0:0: [sdb] Host-managed zoned block device
sd 5:0:0:0: [sdb] 3662151680 4096-byte logical blocks: (15.0 TB/13.6 TiB)
sd 5:0:0:0: [sdb] 55880 zones of 65536 logical blocks
sd 5:0:0:0: [sdb] Write Protect is off
sd 5:0:0:0: [sdb] Mode Sense: 00 3a 00 00
sd 5:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
sd 5:0:0:0: [sdb] Attached SCSI disk
...

The zone model of the disk is confirmed to be "host managed". The total number of zones on the disk is also displayed. In this example, the disk capacity is 15 TB and the disk has 55880 zones.

The zone size of the disk can be inspected by using sysfs to examine the attribute chunk_sectors:

# cat /sys/block/sdb/queue/chunk_sectors 
524288

The value is displayed as a number of 512B sectors regardless of the actual logical and physical block size of the disk. In this example, the disk zone size is 524288 x 512 = 256 MiB.

As of Linux kernel version 4.20.0, the sysfs attribute nr_zones reports the total number of zones on the disk:

# cat /sys/block/sdb/queue/nr_zones
55880

Discovering The Disk Zone Configuration​

To obtain detailed information on the disk zone configuration, for instance the number of conventional zones available, the blkzone utility can be used.

# blkzone report /dev/sdb
  start: 0x000000000, len 0x080000, wptr 0x000000 reset:0 non-seq:0, zcond: 0(nw) [type: 1(CONVENTIONAL)]
  start: 0x000080000, len 0x080000, wptr 0x000000 reset:0 non-seq:0, zcond: 0(nw) [type: 1(CONVENTIONAL)]
  start: 0x000100000, len 0x080000, wptr 0x000000 reset:0 non-seq:0, zcond: 0(nw) [type: 1(CONVENTIONAL)]
  ...
  start: 0x010480000, len 0x080000, wptr 0x000000 reset:0 non-seq:0, zcond: 0(nw) [type: 1(CONVENTIONAL)]
  start: 0x010500000, len 0x080000, wptr 0x000000 reset:0 non-seq:0, zcond: 0(nw) [type: 1(CONVENTIONAL)]
  start: 0x010580000, len 0x080000, wptr 0x000000 reset:0 non-seq:0, zcond: 0(nw) [type: 1(CONVENTIONAL)]
  start: 0x010600000, len 0x080000, wptr 0x000008 reset:0 non-seq:0, zcond: 4(cl) [type: 2(SEQ_WRITE_REQUIRED)]
  start: 0x010680000, len 0x080000, wptr 0x000000 reset:0 non-seq:0, zcond: 1(em) [type: 2(SEQ_WRITE_REQUIRED)]
  start: 0x010700000, len 0x080000, wptr 0x000000 reset:0 non-seq:0, zcond: 1(em) [type: 2(SEQ_WRITE_REQUIRED)]
  ...
  start: 0x6d2280000, len 0x080000, wptr 0x000000 reset:0 non-seq:0, zcond: 1(em) [type: 2(SEQ_WRITE_REQUIRED)]
  start: 0x6d2300000, len 0x080000, wptr 0x000000 reset:0 non-seq:0, zcond: 1(em) [type: 2(SEQ_WRITE_REQUIRED)]
  start: 0x6d2380000, len 0x080000, wptr 0x000000 reset:0 non-seq:0, zcond: 1(em) [type: 2(SEQ_WRITE_REQUIRED)]

This output shows that that the 512B sector range (from 0 up to 0x010600000) is divided into 524 conventional zones. The sector space starting at 0x010600000 and ending at the last sector of the disk is divided into 55356 sequential write required zones.

The zbc_report_zones of libzbc provides more detailed information in a more readable format.

# zbc_report_zones /dev/sdb
Device /dev/sdb:
    Vendor ID: ATA HGST HSH721415AL T220
    Zoned block device interface, Host-managed zone model
    29297213440 512-bytes sectors
    3662151680 logical blocks of 4096 B
    3662151680 physical blocks of 4096 B
    15000.173 GB capacity
    Read commands are unrestricted
    Maximum number of open sequential write required zones: 128
    55880 zones from 0, reporting option 0x00
55880 / 55880 zones:
Zone 00000: type 0x1 (Conventional), cond 0x0 (Not-write-pointer), sector 0, 524288 sectors
Zone 00001: type 0x1 (Conventional), cond 0x0 (Not-write-pointer), sector 524288, 524288 sectors
Zone 00002: type 0x1 (Conventional), cond 0x0 (Not-write-pointer), sector 1048576, 524288 sectors
...
Zone 00521: type 0x1 (Conventional), cond 0x0 (Not-write-pointer), sector 273154048, 524288 sectors
Zone 00522: type 0x1 (Conventional), cond 0x0 (Not-write-pointer), sector 273678336, 524288 sectors
Zone 00523: type 0x1 (Conventional), cond 0x0 (Not-write-pointer), sector 274202624, 524288 sectors
Zone 00524: type 0x2 (Sequential-write-required), cond 0x4 (Closed), reset recommended 0, non_seq 0, sector 274726912, 524288 sectors, wp 274726920
Zone 00525: type 0x2 (Sequential-write-required), cond 0x1 (Empty), reset recommended 0, non_seq 0, sector 275251200, 524288 sectors, wp 275251200
Zone 00526: type 0x2 (Sequential-write-required), cond 0x1 (Empty), reset recommended 0, non_seq 0, sector 275775488, 524288 sectors, wp 275775488
...
Zone 55877: type 0x2 (Sequential-write-required), cond 0x1 (Empty), reset recommended 0, non_seq 0, sector 29295640576, 524288 sectors, wp 29295640576
Zone 55878: type 0x2 (Sequential-write-required), cond 0x1 (Empty), reset recommended 0, non_seq 0, sector 29296164864, 524288 sectors, wp 29296164864
Zone 55879: type 0x2 (Sequential-write-required), cond 0x1 (Empty), reset recommended 0, non_seq 0, sector 29296689152, 524288 sectors, wp 29296689152

Using a SAS Host Bus Adapter​

AHCI adapters can only accomodate Serial ATA disks and generally only provide a limited number of ports. SAS Host Bus Adapters (HBA) are widely used in enterprise applications to overcome AHCI limitations. The SAS transport layer used by SAS HBAs can equally accomodate both Serial ATA and SCSI disks.

HBA Compatibility​

While most AHCI adapters for Serial ATA disks generally do not cause any problem with host managed ZAC disks identification, SAS HBAs on the other hand may suffer from a lack of support depending on the HBA model being used.

The compatibility of a SAS HBA with host managed disks mainly depends on the following factors.

  1. The HBA must have the ability to recognize the host managed device type 0x14 of host managed SAS disks (ZBC/SCSI).

  2. The HBA must have the ability to recognize the host managed device signature 0xabcd of SATA host managed ZAC disks and translate this signature into the ZBC defined SCSI device type 0x14.

  3. Generalizing the previous point, the HBA must implement a SCSI-to-ATA translation (SAT) layer supporting the conversion of host issued ZBC zone commands into ZAC zone commands that can be executed by a SATA ZAC disk connected to the HBA.

Any HBA failing the first requirement will not expose a ZBC host managed disk to the host. Similarly, an HBA failing to comply with the second and third requirement will fail to expose to the host a host managed ZAC disk as a ZBC host managed disk.

In the case of a host aware disk model, the device type and device signature handling will not cause any problem (recall that host aware disks use the regular disk device type and signatur 0x00). Host aware disks will thus always be useable as regular disks with any HBA. The execution of ZBC zone commands with a SAS host aware disk may also work most of the time. However, similarly to host managed disk, the absence of a ZBC/ZAC compatible SAT layer will prevent the use of a Serial ATA host aware disk as a ZBC host aware disk. The ZBC zone commands sent to the SATA disk will not be translated and result in command failures.

The compatibility of an HBA model with the ZBC and ZAC standards should be checked with the HBA vendor. Under some conditions, an HBA compatibility can also be checked using the libzbc conformance test suite.

Verifying The Disk​

If you are using a compatible HBA and you have connected the drive and rebooted the system, verifying the disk identification and checking the disk parameters and zone configuration can be done in the exact same manner as with Serial ATA disks as discussed above.

note

Remember that most SAS HBAs have plug-and-play features that make it unnecessary to reboot the system after connecting or disconnecting a disk to it.

In these examples, /dev/sdc is an SAS disk connected to an SAS HBA and /dev/sdd is a SATA disk connected to the same HBA.

# lsscsi -g
[2:0:0:0]    disk    ATA      INTEL SSDSC2CT18 335u  /dev/sda   /dev/sg0
[5:0:0:0]    zbc     ATA      HGST HSH721415AL T220  /dev/sdb   /dev/sg1
[10:0:2:0]   zbc     HGST     HSH721414AL52M0  a220  /dev/sdc   /dev/sg2
[10:0:3:0]   zbc     ATA      HGST HSH721415AL T220  /dev/sdd   /dev/sg3

Inspecting the kernel messages shows no differences between the initialization of the SAS drive and the SATA disk.

# dmesg
...
scsi 10:0:2:0: Direct-Access-ZBC HGST     HSH721414AL52M0  a220 PQ: 0 ANSI: 7
scsi 10:0:2:0: SSP: handle(0x001b), sas_addr(0x5000cca0000025c5), phy(1), device_name(0x5000cca0000025c7)
scsi 10:0:2:0: enclosure logical id (0x500062b200f35d40), slot(2)
scsi 10:0:2:0: enclosure level(0x0000), connector name(     )
sd 10:0:2:0: Attached scsi generic sg2 type 20
sd 10:0:2:0: [sdc] Host-managed zoned block device
sd 10:0:2:0: [sdc] 27344764928 512-byte logical blocks: (14.0 TB/12.7 TiB)
sd 10:0:2:0: [sdc] 4096-byte physical blocks
sd 10:0:2:0: [sdc] 52156 zones of 524288 logical blocks
sd 10:0:2:0: [sdc] Write Protect is off
sd 10:0:2:0: [sdc] Mode Sense: f7 00 10 08
sd 10:0:2:0: [sdc] Write cache: enabled, read cache: enabled, supports DPO and FUA
sd 10:0:2:0: [sdc] Attached SCSI disk
...
scsi 10:0:3:0: Direct-Access-ZBC ATA      HGST HSH721415AL T220 PQ: 0 ANSI: 6
scsi 10:0:3:0: SATA: handle(0x001c), sas_addr(0x300062b200f35d43), phy(3), device_name(0x5000cca25bc2e26f)
scsi 10:0:3:0: enclosure logical id (0x500062b200f35d40), slot(1)
scsi 10:0:3:0: enclosure level(0x0000), connector name(     )
scsi 10:0:3:0: atapi(n), ncq(y), asyn_notify(n), smart(y), fua(y), sw_preserve(y)
sd 10:0:3:0: Attached scsi generic sg3 type 20
sd 10:0:3:0: [sdd] Host-managed zoned block device
sd 10:0:3:0: [sdd] 3662151680 4096-byte logical blocks: (15.0 TB/13.6 TiB)
sd 10:0:3:0: [sdd] 55880 zones of 65536 logical blocks
sd 10:0:3:0: [sdd] Write Protect is off
sd 10:0:3:0: [sdd] Mode Sense: 9b 00 10 08
sd 10:0:3:0: [sdd] Write cache: enabled, read cache: enabled, supports DPO and FUA
sd 10:0:3:0: [sdd] Attached SCSI disk
...

Both disks are identified by the kernel as Direct-Access-ZBC devices. This indicates that the HBA is correctly translating the ZAC host managed device signature into a ZBC host managed device type.