Pulsar XT.2 SAS Product Manual, Rev. B
15
6.1.1
Unrecoverable Errors
An unrecoverable data error is defined as a failure of the drive to recover data from the media. These errors
occur due to read or write problems. Unrecoverable data errors are only detected during read operations, but
not caused by the read. If an unrecoverable data error is detected, a MEDIUM ERROR
(
03h) in the Sense Key
will be reported. Multiple unrecoverable data errors resulting from the same cause are treated as 1 error.
6.1.2
Interface errors
An interface error is defined as a failure of the receiver on a port to recover the data as transmitted by the
device port connected to the receiver. The error may be detected as a running disparity error, illegal code, loss
of word sync, or CRC error.
6.2
Endurance Management
Customer satisfaction with Solid State Drives can be directly related to the internal algorithms which an SSD
uses to manage the limited number of Program-Erase (PE) cycles that NAND Flash can withstand. These
algorithms consist of Wearleveling, Garbage Collection, Write Amplification, Unmap, Data Retention, Lifetime
Endurance Management.
6.2.1
Wear Leveling
Wear Leveling is a technique used by the drive to ensure that all Flash cells are written to or exercised as
evenly as possible to avoid any hot spots where some cells are used up faster than other locations. Wear Lev-
eling is automatically managed by the drive and requires no user interaction. The Seagate algorithm is tuned to
operate only when needed to ensure reliable product operation.
6.2.2
Garbage Collection
Garbage Collection is a technique used by the drive to consolidate valid user data into a common cell range
freeing up unused or obsolete locations to be erased and used for future storage needs. Garbage Collection is
automatically managed by the drive and requires no user interaction. The Seagate algorithm is tuned to oper-
ate only when needed to ensure reliable product operation.
6.2.3
Write Amplification
While Write Amplification is not an algorithm, it is a major characteristic of SSD's that must be accounted for by
all the algorithms that the SSD implements. The Write Amplification Factor of an SSD is defined as the ratio of
Host/User data requested to be written to the actual amount of data written by the SSD internal to account for
the user data and the housekeeping activities such as Wear Leveling and Garbage Collection. The Write
Amplification Factor of an SSD can also be directly affected by the characteristics of the host data being sent to
the SSD to write. The best Write Amplification Factor is achieved for data that is written in sequential LBA's that
are aligned on 4KB boundaries. The worst case Write Amplification Factor typically occurs for randomly written
LBA's of transfer sizes that are less than 4KB and that originate on LBA's that are not on 4KB boundaries.
6.2.4
UNMAP
A new SCSI command has been added to the SSD as part of the Thin Provisioning feature set. Use of the
UNMAP command reduces the Write Amplification Factor of the drive during housekeeping tasks such as
Wear Leveling and Garbage Collection. This is accomplished because the drive does not need to retain data
which has been classified by the host as obsolete.
