If you've got data, post it. If not, it starts to sound like astroturf.
Fwiw, with my small sample size, I agree with backblaze. Seagates appear to be failing more, and avoid wdc greens.
> and avoid wdc greens
Though that may just be my faulty memory of he The original piece.
I believe the backblaze design has 1 pound drive vibrating at 120hz or so supported by a SATA connector. I don't know the SATA specification, but I suspect they aren't designed to support the drives weight. If SATA connectors where designed to support the drive they wouldn't be plastic.
Additionally the SATA Multiplier boards (thin fiberglass with delicate traces) are supporting the weight of 5 drives each.
This would also explain why the enterprise drives (typically heavier, more platters, and with a more conservative design) are less reliable then the consumer (with typically higher density platters).
Also, insisting that Backblaze has some ridiculously out-of-this-world "workload"? We're not talking a database, but an operation that is usually done with tapes. According to him, consumers must only need 4TB drives with 1000 hour MTBFs and 20TB of lifetime IOs.
Backblaze already acknowledged that their usage environment doesn't reflect that of normal consumer, yet the author harps on the point.
The author tries to debunk Backblaze's claim that it found no correlation between drive temperature and failure rate. Yet he presents no evidence that Backblaze's methodolgy was wrong.
Backblaze had large enough sample to have even distribution of drives in various temperature, rack, and load environment. I doubt they put all Seagate drives in the middle shelf of Ver. 1 racks where it was the hottest, but put all Hitachi drives in the bottom of the Ver. 3 racks where it was cooler and more stable.
Unless there is another study that shows failure rate of these drives in normal consumer environment, Seagate's 3-5x failure rate should give people a pause.