Low-cost cloud backup and storage firm Backblaze has revealed its newest set of laborious disk reliability numbers for the second quarter of 2017. Whereas the corporate has tended to stay with consumer-oriented laborious disks, a great pricing deal has meant that it additionally now has a number of thousand enterprise-class disks, permitting for some large-scale comparisons to be drawn between the 2 sorts of storage. The corporate has additionally began to accumulate bigger disks with capacities of 10TB and 12TB.
The corporate is utilizing two fashions of 8TB Seagate disk; one client, with a two-year guarantee, the opposite enterprise, with a five-year guarantee. Final quarter, Backblaze famous some efficiency and energy administration benefits to the enterprise disks, however for the corporate’s important use case, these have been of considerably marginal worth. The efficiency does assist with preliminary knowledge migrations and ingest, however the efficiency profit general is restricted, as a result of manner Backblaze distributes knowledge over so many spindles.
In mixture, the corporate has now collected three.7 million drive days for the buyer disks and 1.four million for the enterprise ones. Over this utilization, the annualized failure charges are 1.1 % for the buyer disks, and 1.2 % for the enterprise ones. No less than for now, then, the enterprise disks aren’t doing something to justify their longer guarantee; their reliability is nearly an identical. The main focus now could be on what occurs to the buyer disks as they cross their two-year guarantee interval. Will they present the identical reliability, or will deterioration change into extra obvious?
Historically, the anticipated sample of laborious disk failures has been the so-called “bathtub curve:” a spate of failures of recent disks (“toddler mortality”), as disks that have been faulty from the manufacturing unit are shipped into the wild and quickly cease working, adopted by a protracted interval of low failure charges, after which an uptick in failures because the disks cross their engineered lifetime.
The preliminary knowledge from the 10TB and 12TB disks, nevertheless, has not proven that sample. Whereas the information to date could be very restricted, with 1,240 disks and 14,220 mixture drive days collected to date, none of those disks (each Seagate fashions) have failed. The low stage of utilization implies that the disks have been put in and formatted and never a lot past that, however true toddler mortality—disks that instantly expire on their first use—hasn’t change into obvious. After all, it is too early to know if this pattern will proceed long-term, however for 1,240 disks to be purchased and put in with out a single one failing to spin up or having another defect is nonetheless uncommon.