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 ...