SSD DC S3700 Wearout Measuring

[rmare2]

For our product, we are developing with DC S3700 series "INTEL SSDSC2BA400G3" SSDs and we are trying to understand the life span. From the SMART data we get conflicting numbers from the "Total LBAs Written (0xf1)" and "Media Wear out Indicator (0xE9)."

Could you help us understand how to determine SSD Life? And why the two SMART parameters tell a different story?

Intel Spec:

400 GB SSDs should have a life of 7.30 PBW

SMART Data:

"Total LBAs Written": 0xf1 : 100: 100: RAW: 0x00000cc41bab

This translates to 7.18 PB written = (214178731 count)*(65536 sectors/count)*(512 bytes/sector)/(1000^5).

Though, the media wear smart parameter shows 55

"Media Wearout Indicator": 0xe9 : 55: 55: RAW: 0x000000000000

So, why do the "Media Wearout Indicator" and the "Total LBAs Written" when using the drive specification limit differ so greatly?

[jbenavides]

Hello wlrm04,

We are checking if there has been any progress regarding your inquiry. We will keep you informed as soon as we have an update.

[jbenavides]

Please let us know the method you are using to get 0xe9 value.

- Application name and version. For example: HDParm, SmarmonTools, or other.

- Operating System

Have you noticed any difference in the values since the original post?

[rmare2]

----------------------------------------------------------------------------

CrystalDiskInfo 5.6.2 (C) 2008-2013 hiyohiyo Crystal Dew World : http://crystalmark.info/ http://crystalmark.info/----------------------------------------------------------------------------

OS : Windows 7 Professional SP1 [6.1 Build 7601] (x64)

Date : 2015/01/16 11:40:23

----------------------------------------------------------------------------

-- S.M.A.R.T. --------------------------------------------------------------ID Cur Wor Thr RawValues(6) Attribute Name05 100 100 __0 000000000000 Re-Allocated Sector Count09 100 100 __0 000000001FD0 Power-On Hours Count0C 100 100 __0 000000000B75 Power Cycle CountAA 100 100 _10 000000000000 Available Reserved SpaceAB 100 100 __0 000000000000 Program Fail CountAC 100 100 __0 000000000000 Erase Fail CountAE 100 100 __0 000000000B53 Unexpected Power LossAF 100 100 _10 001D00000298 Power Loss Protection FailureB7 100 100 __0 000000000291 SATA Downshift CountB8 100 100 _90 000000000000 End to End Error Detection CountBB 100 100 __0 000000000000 Uncorrectable Error CountBE _86 _86 __0 00000E0D000E TemperatureC0 100 100 __0 000000000B53 Unsafe Shutdown CountC2 100 100 __0 00000000000E TemperatureC5 100 100 __0 000000000000 Pending Sector CountC7 100 100 __0 000000000F9C CRC Error CountE1 100 100 __0 00000CC5236A Host WritesE2 100 100 __0 00000000FFFF Timed Workload Media WearE3 100 100 __0 0000FFFFFFFF Timed Workload Host Read/Write RatioE4 100 100 __0 00000000FFFF Timed Workload TimerE8 100 100 _10 000000000000 Available Reserved SpaceE9 _55 _55 __0 000000000000 Media Wearout IndicatorEA 100 100 __0 000000000000 Vendor SpecificF1 100 100 __0 00000CC5236A Total LBAs WrittenF2 100 100 __0 00000016F8E0 Total LBAs Read

[jbenavides]

Hello wlrm04,

We would like to provide the following information about this topic:

The Media Wearout Indicator and the Endurance ratings do not have a direct relation. The E9h SMART attribute will not match wear calculation based on the LBA's written.

- The reliability values in the product specifications are a minimum expected for Intel® Solid-State Drives.

The Endurance rating is calculated using Global standards. It is defined using JESD218 testing method and and JESD219 workloads. However, we expect our drives to exceed those values in almost all cases.

- The Media Wearout Indicator is calculated by comparing the actual number of Program/Erase (P/E) cycles experienced by the NAND chips, to the maximum number of P/E cycles that the chips are designed to endure. It does not compare the amount of bytes written to the Endurance values in the Spec sheet.

This makes the Media Wearout Indicator a true indicator of the wear of the chips.

[rmare2]

Thank you for your reply on this topic.

Do you have a recommendation as to how we can deterministically have our systems warn us based on our write rate when a drive will fail in a month?

The "Media Wearout" indicator does not seem to give the user the ability to estimate how much more the user can use the SSD before the indicator is 1.

For our product, we liked the ability to use the written count so that we could have a user deterministic method to say our product with Intel SSDs should be able to last about X amount of time based on current SMART status and our write rate. Of course, we do not want to replace SSDs that are still good.