Posted May 19, 2016
One billion hours on, and HGST still rules the roost for hard disk reliability
Even five-year-old disks are still going strong.
Cloud backup provider Backblaze has published the latest data it has accumulated about the reliability of the hard drives it uses. In the first quarter of the year, the company passed more than a billion hours of aggregate drive usage since it started tracking reliability in April 2013.
HGST's drives have long stood out as the most reliable, and that trend continues. Their failure rate is remarkably low; even after three years in service, the 3TB and 4TB units have annualized failure rates of just 0.81 percent and 1.03 percent, respectively. 2TB units, which last quarter were already on average more than 5 years old, have seen a small increase in failure rate—1.57 percent, compared to 1.15 percent a year ago—but still show extraordinary reliability considering their age.
After some bad experiences with certain models and annualized failure rates in some cases approaching 30 percent, Seagate's performance is also solid. Backblaze's most common disk type is a 4TB Seagate unit, with nearly 35,000 of the drives in use, and those are demonstrating at a failure rate of 2.90 percent.
The company continues to substantially stick with Seagate 4TB units, in spite of somewhat worse failure rates, due to a combination of better pricing and availability. Backblaze says that it typically orders disks 5,000 to 10,000 at a time, and while it has found suppliers of Seagate and (Western Digital-owned) HGST that can handle these orders, it has struggled to do so consistently for Western Digital and Toshiba disks. This availability concern also pushes the company toward 4TB units over 6 or 8TB ones; although the pricing of those is starting to make them cost-effective, their bulk availability is still limited.
While the company uses a mixture of different disks, within each of its Vaults (systems of 20 individual Storage Pods, with older storage pods holding 45 drives and the latest ones increasing that to 60) it standardizes on a particular type, so Backblaze needs to be able to buy 1,200 disks at once to be able to deploy disks at any kind of a reasonable scale.
source
I bought recently (months ago) a 1 TB WD unit. Fingers crossed ^_^
Even five-year-old disks are still going strong.
Cloud backup provider Backblaze has published the latest data it has accumulated about the reliability of the hard drives it uses. In the first quarter of the year, the company passed more than a billion hours of aggregate drive usage since it started tracking reliability in April 2013.
HGST's drives have long stood out as the most reliable, and that trend continues. Their failure rate is remarkably low; even after three years in service, the 3TB and 4TB units have annualized failure rates of just 0.81 percent and 1.03 percent, respectively. 2TB units, which last quarter were already on average more than 5 years old, have seen a small increase in failure rate—1.57 percent, compared to 1.15 percent a year ago—but still show extraordinary reliability considering their age.
After some bad experiences with certain models and annualized failure rates in some cases approaching 30 percent, Seagate's performance is also solid. Backblaze's most common disk type is a 4TB Seagate unit, with nearly 35,000 of the drives in use, and those are demonstrating at a failure rate of 2.90 percent.
The company continues to substantially stick with Seagate 4TB units, in spite of somewhat worse failure rates, due to a combination of better pricing and availability. Backblaze says that it typically orders disks 5,000 to 10,000 at a time, and while it has found suppliers of Seagate and (Western Digital-owned) HGST that can handle these orders, it has struggled to do so consistently for Western Digital and Toshiba disks. This availability concern also pushes the company toward 4TB units over 6 or 8TB ones; although the pricing of those is starting to make them cost-effective, their bulk availability is still limited.
While the company uses a mixture of different disks, within each of its Vaults (systems of 20 individual Storage Pods, with older storage pods holding 45 drives and the latest ones increasing that to 60) it standardizes on a particular type, so Backblaze needs to be able to buy 1,200 disks at once to be able to deploy disks at any kind of a reasonable scale.
source
I bought recently (months ago) a 1 TB WD unit. Fingers crossed ^_^