I am collecting the SMART data from a DC S3610 series SSD drive and want to set an alarm when the data goes out of tolerance. If I am collecting the raw instead of normalized data what would the thresholds be to predict drive failure? Another way to come up with those thresholds would be to use the equations that turn the raw data into the normalized data, then I could choose a normalized threshold and calculate a raw value threshold. Are those equations available?
Most of this information is readily available in the Product Specification document for each solid state drive model. It's important to keep in mind that each SMART attribute may collect a different type of data, and each calculation may need to be adjusted depending on the drive size.
- http://www.intel.com/content/dam/www/public/us/en/documents/product-specifications/ssd-dc-s3610-spec... Intel® Solid-State Drive DC S3610 Series Product Specification.
Example: From Section 5.4.1, Table 19: SMART Attributes.
- Attribute: Total Bytes Written
"The raw value of this attribute reports the total number of sectors written to the NAND media.
This includes NAND writes triggered by host writes, defrag, background data refresh and
wear level relocation writes etc. The raw value is increased by 1 for every 65,536 sectors (32MB)
writes to the NAND media. Upon NAND write, new value returned once per minute."
- Flags: Self-preserving attribute, Event count attribute, Online collection attribute. Not an error rate attribute, Not a performance attribute, Advisory.
- Threshold: 0 (none)
Please let us know if this is what you were looking for. If not, we'll reach out to our additional resources and see what we can come up with.