Cloud storage company Backblaze offers a great deal of transparency into their deployments. Here's what you need to know before buying your next hard drive.
For years, Backblaze has published reports detailing the drives deployed in their "Storage Pods" used for backing up servers and client computers into the cloud. Typically, enterprise-scale cloud storage vendors use drives labeled explicitly for enterprise use, though Backblaze carried the distinction of using the least expensive drives they could acquire—at times, shucking external hard drives found on sale, especially in the aftermath of 2012 flooding in Thailand severely disrupting the drive market.
The Q2 2019 BlackBlaze Hard Drive Stats report is a largely surprise-free affair—of the 108,461 drives in service, 474 failed during the second quarter, of which 446 were Seagate. It's not a bad showing, particularly in comparison to prior years, though it is disproportionate. In total, 74.3% of drives deployed by Backblaze are Seagate products, with Seagate reporesenting 94.1% of the failed drives.
The remaining 5.9% of failed drives are HGST-branded products, with HGST representing 24.5% of Backblaze's deployment. No Toshiba drives failed in Q2, though Toshiba only represents 1.2% of Backblaze's deployment, making it statistically insignificant. Backblaze is evaluating new Western Digital drives, though presently does not have any in mass deployment.
Looking at individual models of failed drives, 52% are Seagate Exos 12TB ST12000NM0007, an enterprise-targeted, helium-sealed drive with eight platters spinning at 7200 RPM. Since first deployment, 737 of these have failed—247 of those failures happened this quarter. Drive age is unlikely to be a significant factor here, as Backblaze lists the average age as 13.58 months. The annualized failure rate (AFR) is 1.89%, the second-highest of all the drives in deployment.
SEE: Top five on-premises cloud storage options (free PDF) (TechRepublic)
Likewise, 22% of those that failed are desktop-focused Seagate 4TB ST4000DM000 drives. These are relatively low-speed, budget drives spinning at 5900 RPM, and are not touted for enterprise storage use cases. These are among the oldest drives that Backblaze has in service, with an average age of 44.41 months, and are the most prone to failure—of the 19,570 deployed, 3,652 have failed, with an annualized failure rate of 2.72%.
Avoiding the 12TB Seagate Exos drive due to high failure rates is a reasonable and defensible caution, while avoiding the desktop-class 4TB Seagate drives for your next build is just a good idea—Backblaze noted in the Q1 2019 report that they plan to decommission over half their 4TB drives by the end of the year. That said, Seagate prices do often undercut the competition, and are certainly worth consideration.
Conversely, the lack of failures from Toshiba drives for the quarter should not be interpreted as a ringing endorsement, as the number of Toshiba drives in service makes that statistic less than meaningful. The lowest AFR award for all drives to date (not just the last quarter) goes to the HGST Ultrastar DC HC520 (HUH721212ALN604), with only 0.37%, representing 10 failed drives out of 9,609 deployed. Despite that statistic, caution is still warranted—these are the newest drives in deployment, with an average age of 3.52 months.
As always, the biggest takeaway is back up your data regularly, and test your backups.
For more, check out "Increasing storage densities require aggressive software management to maintain performance" and "Seagate preparing dual-actuator hard drive for data centers: Will it work?" on TechRepublic.
- Hyperconverged infrastructure: A cheat sheet (TechRepublic)
- 10 things companies are keeping in their data centersr (TechRepublic download)
- Hiring kit: Database administrator (TechRepublic Premium)
- The data center is dead: Here's what comes next (ZDNet)
- Best cloud services for small businesses (CNET)
- DevOps: More must-read coverage (TechRepublic on Flipboard)