Viking Technology Technology Computer Drive PSF1XXXXGXXXX ユーザーズマニュアル
Manual
1/2/2014
PSFS1XXXXGXXXX
Viking Technology
Revision B7
Page 22 of 60
www.vikingtechnology.com
2.6 Reliability
Table 2-12: Reliability Specifications
Parameter
Value
Nonrecoverable read errors (UBER)
1
<1 sector in 10
17
bits read, max
Mean Time Between Failures (MTBF)
2
3,000,000 hours
Power On/Off Cycles
3
50,000 cycles
Read Endurance
Unlimited
Write or Erase Endurance
4
(specified by the flash component)
Global wear-leveling
~ 2% between least worn and most worn cell
Data retention
See Note 5
Notes:
1. Uncorrectable Bit Error rate (UBER) will not exceed one sector in the specified number of bits read. In the
1. Uncorrectable Bit Error rate (UBER) will not exceed one sector in the specified number of bits read. In the
extremely unlikely event of a non-recoverable read error, the drive will report it as a read failure to the host;
the sector in error is considered corrupt and is not returned to the host.
the sector in error is considered corrupt and is not returned to the host.
2. MTBF is calculated based on a Part Stress Analysis. It assumes nominal voltage, with all other parameters
within specified range. Telcordia method SR-332, component FIT rate at 55
o
c.
3. Power On/Off Cycles defined as power being removed from the drive, and then restored. Note that host
systems and drive enclosures may remove power from the drive for reasons other than a system
shutdown.
shutdown.
4. SLC NAND has a higher write/erase cell endurance than MLC NAND; varies by NAND component.
5. 10 years for P-E cycle counts < 10% of NAND spec and 1 year for max-spec P-E cycle counts per JEDEC-
5. 10 years for P-E cycle counts < 10% of NAND spec and 1 year for max-spec P-E cycle counts per JEDEC-
STD-JESD47) Refer to Viking Application Note AN0011 – Flash Data Retention
2.6.1 Data, MetaData, and Firmware Code Protection
Element SSD implements data protection throughout its data path. Protection
techniques include:
techniques include:
• Data ECC Algorithms
• Datapath CRC Error Detection
• RAISE
TM
Data Protection Against Catastrophic Flash Page/Block Failure
2.6.1.1 DATA ECC Algorithms
The following data error correction is provided:
• For Flash memory devices providing 128 bytes of redundancy per 4k of
data (normally this is SLC Flash)
o 16 bytes of redundancy applied to 512 bytes of data
o Up to seven 9-bit symbols (up to 63 bits if contiguous) correctable
o 16 bytes of redundancy applied to 512 bytes of data
o Up to seven 9-bit symbols (up to 63 bits if contiguous) correctable
• For Flash memory devices providing 218 or more bytes of redundancy per
4k of data (normally this is MLC Flash)
o 27 bytes of redundancy applied to 512 bytes of data
o Up to twelve 9-bit symbols (up to 108 bits if contiguous) correctable
o 27 bytes of redundancy applied to 512 bytes of data
o Up to twelve 9-bit symbols (up to 108 bits if contiguous) correctable