Methodology Backblaze, which open sourced their Storage Pod a few years ago, is now giving drive failure rates. They currently have over 27,000 consumer grade drives spinning in Backblaze storage pods.
Almost 13,000 each are Seagate and Hitachi drives, almost 3000 Western Digital drives and a too small for statistical reporting smattering of Toshiba and Samsung drives.
One cool thing: Backblaze buys drives the way you and I do: they get the cheapest drives that will work. Their workload is almost hundred percent write. Because they spread the incoming writes over several drives their workload isn't very performance intensive either.
Also, like you and me, they are willing to spend a little more for a more reliable drive but not a lot more. And unlike you and me they track every single drive that they install.
They measure the annual failure rate (AFR) rather than MTBF because that's an easier number to understand. They count as a failure anytime they have to replace a drive. Of course, vendors report that about half of all returned drives have no trouble found – but I like Backblaze's definition better.
The bad news Annual failure rates are all over the map ranging from as low as 0.9% to as high as 25.4% - and one drive hit 120%. But the huge majority of drives that Backblaze has installed over the last four years are still operating.
Backblaze excluded some drives from the results because they just don't work at all in the Backblaze environment. These are some of the so-called green drives – WD's 3TB Green drives and Seagate's LP (low-power) 2TB drives.
Backblaze is not equipped to do root cause analysis on drive failures so they don't know the reasons for the problems. Other Seagate LP drives have been quite reliable, which points up the limits of blanket assumptions about an entire model family.
Hall of shame In general, Backblaze sees more problems with Seagate drives than the other two vendors. Yet those problems have been getting better with each successive generation of high-capacity drives.
But the 1.5 TB Seagate Barracuda 7200 and the Seagate Barracuda Green have the highest AFRs, 25.4% and 120% respectively. And in general the Seagate drives have proven to be less reliable overall.
Hall of Fame Hitachi drives have proven to be the most reliable. Their annual failure rates have ranged from .9% all the way up to 1.5%.
Of course Hitachi drives typically cost more. Maybe you do get what you pay for.
In the middle are the Western Digital drives with AFRs ranging from 3.2% to 7.3%. Of course, WD will take full control of Hitachi in a few months and I suspect that we will see Hitachi's failure rates approximate those of WD over time.
Here's the Backblaze table of results:
The Storage Bits take All large drive buyers and the vendors track reliability. But long ago the industry decided not to share this information with consumers.
Backblaze is to be commended for breaking the industry-wide code of silence. Our digital data is too precious for consumers to be left in the dark.
You may wonder why Hitachi who has long had a reputation for good quality is the company that was acquired. Industry scuttlebutt has it that Hitachi's yields - the percentage of shippable drives compared to starts – was the lowest in the industry. This left them with high costs and their profitability suffered. People just weren't willing to pay for quality.
It is encouraging that Seagate's 4TB drive AFRs are more in line with the competition. Perhaps they had some manufacturing hiccups with earlier drives. And remember, these are 3.5" drives: notebook drives may be different.
But perhaps, like Backblaze, you will choose to buy the lowest cost drive that offers reasonable reliability. Backblaze still buys a lot of Seagate drives and there is nothing in this report that would cause me not to buy Seagate either.
But for my RAID array? Yeah, I like Hitachi.
The bottom line is that there is no substitute for backing up your data. Even in the most reliable drives do fail and if you haven't back up your data it is you - not the vendor - who is the loser.
Comments welcome. Are there any surprises in this data for you? Read the entire Backblaze blog post.