S.M.A.R.T. attributes description

The generic description

The S.M.A.R.T. attributes are the specific properties that are used for analysis of the state and the productivity reserve of the disk. HDD manufacturer chooses the list of these attributes according to their capacities, in order to predict the deterioration of the operating characteristics or to determine HDD's defectiveness. Every manufacturer has their own list of attributes and can change it freely, according to new requirements, without a notice.

The attribute values

The attribute's value is used to present relative reliability of an operational or etalon characteristic. The allowable value can lie in the range from 1 to 255 (as a rule from 1 to 100, but sometimes from 1 to 200). A higher value indicates that the result of the analysis of this operational characteristic signifies a low probability of its degradation or HDD failure and vice versa.

The threshold values

Every attribute has its own threshold value, which is used for the comparison with the attribute's value. If the attribute's value is less or equals its threshold value, then it can indicate the deterioration of the operating characteristics or HDD's defectiveness and S.M.A.R.T considers the disk to be faulty. The threshold value is determined by HDD manufacturer by means of the constructive characteristics of the HDD and on the basis of analysis of the reliability tests. The threshold value indicates the bottom possible boundary of the attribute's value. The allowable threshold value can lie in the range from 1 to 255.

The list of the S.M.A.R.T. attributes

The attributes' assignation and the method of interpretation of their values were obtained experimentally or were taken in the support services of the HDDsÆ manufacturers.

Summary table with all of the attributes we know (54) and short descriptions for most of them (45) follows.

The short description of the known S.M.A.R.T. attributes

* Raw Read Error Rate

is the read error rate during read operations from the disk surface, due to a HDD's hardware component failure.

* Throughput Performance
is the average performance (carrying capacity) of the HDD.

A decrease of the attribute's value signifies a great probability that the HDD has some problems.

* Spin Up Time

is the average spin-up time from 0 RPM to the operational rotation speed (5400, 7200 or higher RPM).

Admittedly, the raw value contains the spin up time in milliseconds/seconds.

* Start/Stop Count

is the count of start/stop cycles of the HDD's spindle.

The raw value contains the total start/stop count.

* Reallocated Sectors Count
When HDD encounters a read/write/verification error it tries to reallocate the data from the "bad" sector to a special spare area. If the reallocation operation (also called remapping) was successful, then this sector is marked as an "remapped". Due to the remapping, you will rarely find a bad block during the surface test for a modern HDD.

The raw value contains the total reallocated sectors count.

* Read Channel Margin

reserve of channel while reading.

* Seek Error Rate

Seek errors occur in cases of mechanical position control system failure, servo damage or considerable disk thermal enlargement, etc. The more seek errors there are, the worse state of the mechanical HDD's part and/or the disk surface is.

* Seek Time Performance
is the average seek time performance on a sector. The decrease of the attribute's value signifies troubles in the mechanical HDD's part.

* Power-On Hours/Minutes/Seconds/Half-Minutes
The raw value contains the number of hours (minutes, seconds or half-minutes; it depends on the manufacturer) the HDD had been worked by the moment. The decrease of the attribute's value down to the threshold value signifies achievement of MTBF (Mean Time Between Failures).

Some measures, like power-on hours or number of contact start/stops are easy to measure, but have no certain limit. They are somewhat like the number of miles on a car's odometer. 100,000 miles is a high amount, but it does not mean that the particular car will fail anytime soon.

* Spin Retry Count
The raw value contains the number of attempts to spin the HDD's spindle up to the operational rotation speed if the first attempt was unsuccessful. The decrease of the attribute's value indicates the troubles in the mechanical HDD's part.

* Calibration Retry Count
number of attempts to calibrate the device. The raw value contains the number of attempts to clean the HDD's state and to set heads to the lead-in area if the first attempt was unsuccessful. The decrease of the attribute's value indicates the troubles in the mechanical HDD's part.

* Device Power Cycle Count
Raw value contains the total number of drive start/stop cycles (including both power on/off switching and suspend/wakeup switching).

* Soft Read Error Rate
frequency of "program" errors appearance while reading data from a disk.

* Emergency Re-track
The raw value contains the total number of the heads' leading out to the special parking zone.

* Temperature
The raw value contains the temperature of the HDD (if it is equipped with a thermal sensor), Celsius.

* Reallocation Event Count
quantity of remapping operations. The raw value contains the total number of the remapping attempts (including not only successful but also abortive attempts).

* Current Pending Sector Count
current quantity of unstable sectors (waiting for remapping). The raw value contains the total number of the sectors that HDD considers to be the pretenders to be remapped to the spare area. If one of them is read successfully, then it will be excluded from the pretenders list. Otherwise it will be remapped. The permanently nonzero raw value of this attribute indicates a poor quality of the disk surface.

* Uncorrectable Sector Count
quantity of uncorrected errors. The raw value contains the total number of sectors that could not be corrected during read/write operations. The increase of raw value indicated either evident disk surface defects and/or troubles with the mechanical HDD's part.

* UltraDMA CRC Error Count
UltraDMA controller performs an error checking on data it receives from HDD, ensuring that data was not damaged while transmission over the cable. Each time the error is detected, the controller requests a retransmission, thus slowing down the overall transfer speed. Lower values of "UltraDMA CRC Error Rate" correspond to higher number of errors, usually indicating a cabling problem.

* Write Error Rate (Multi Zone Error Rate)
Similar to Read Error Rate.

* Disk Shift
is a shift of the disks set relative to the spindle axis. The actual value is stored in the raw value. Measure units are unknown.

Note : disk shift may occur as a result of a strong shock load on the HDD or for other reasons (sometimes: Temperature).

* Temperature (Western Digital)
It stores temperature in Celsius, just like Attribute 194. But it has a failure threshold set to correspond to the maximum design operating temperature of the disk, which is 55 Celsius on the WD800JD drives. So if this Attribute has "failed in the past" this means that the maximum disk operating temperature has been exceeded.

* G-Sense Error Rate
frequency of mistakes appearance as a result of impact loads. The raw value contains the indication of shock sensor about the total error count originated as a result of the shock load to HDD (dropping drive, for example).

* Loaded Hours
number of hours in general operational state . The raw value contains the total number of hours (minutes/seconds) when heads were in the run position.

* Load Friction
loading on magnetic heads actuator caused by friction in mechanical parts of the store. Only the time when heads were in the operating/working position is counted.

* Load-in Time
general time of drive loading. Admittedly, the raw value contains the total working time on conditions that the heads are in the run position (out of the parking zone).

* Power-Off Retract Count
quantity of the fixed turning off's a drive. The raw value contains the total number of attempts to switch HDD's supply on/off.

* GMR Head Amplitude
defines the amplitude of heads trembling (GMR-head) in run mode.

* Hardware ECC Recovered
number of ECC on-the-fly errors.

* Soft Read Error Rate
frequency of "program" errors appearance while reading data from a disk.

* Data Address Mark Errors
number of Data Address Mark (DAM) errors (or) vendor-specific.

* Off Track Errors
frequency of the off track errors (Maxtor: Off Track Errors).

* Run Out Cancel
frequency of the ECC errors (Maxtor: ECC Errors).

* Soft ECC Correction
quantity of errors corrected by software ECC.

* Thermal Asperity Rate
frequency of the thermal asperity errors.

* Flying Height
the height of the disk heads above the disk surface.

* Spin High Current
quantity of used high current to spin up drive.

* Spin Buzz
quantity of used buzz routines to spin up drive.

* Offline Seek Performance
drive's seek performance during offline operations.

* Load/Unload Cycle Count
number of cycles into Landing Zone position.

* Load/Unload Retry Count
loading on drive caused by numerous recurrences of operations like: reading, recording, positioning of heads, etc.

* Torque Amplification Count
quantity efforts of the rotating moment of a drive.

* Head Flying Hours
time while head is positioning.

* Read Error Retry Rate
frequency of errors appearance while reading data from a disk.

* Power-Off Retract Cycle
number of the fixed "turning off drive" cycles (Fujitsu: Emergency Retract Cycle Count).

* Shock Sense Error Rate
number of errors as a result of impact loads as detected by a shock sensor (dropping drive, for example).