Backblaze is one of the internet’s newest and youngest cloud storage providers, having based their business model for their service around the use of cheap, replaceable hard drives in order to cut down on running costs for the business. Every year they take the data they collect on drive failures and running lifetimes and compile them into a report for consumers to look at and analyse. Past reports have shown up issues that some hard drive manufacturers have had with some product lines and it also paints big neon signs around specific drive sizes that may be good enough for general consumer use, but are to be avoided for any mission-critical operations.
Its worth noting that part of Backblaze’s business model also revolves around buying up external hard drives in bulk, ripping them out of their casings and putting them into active duty 24/7. Most of the time they just buy as many consumer drives from the distribution channel as they need. These drives are inserted into the company’s custom-designed storage pods, self-contained units of up to 45 drives that could potentially hold up to 270 terabytes of data.
The storage pods do not rely on any form of RAID for data security, so Backblaze instead uses custom software to perform data duplication across multiple pods in their data center. As of the most recent report compiled for 2014, Backblaze has 41,213 drives in operation in their data center.
Overall, Seagate’s 3TB consumer drives fared the worst. They didn’t last as long under Backblaze’s control standards and they failed way more often in 2014. The 1.5TB drives suffer just as much, but to a lesser degree, probably because those are two-platter drives. Looking at how the storage pods are arranged, its a wonder that the other 55-odd percent of 3TB Seagate drives survived this year at all.
Its also worth pointing out that one of Backblaze’s strategies is to replace storage pods regularly with drives of higher capacity, so most of the pods containing 1.5TB drives would have simply been replaced with 3TB drives. After a pod migration, if any drives are still fully operational and perform to their standard, Backblaze takes them out of the reliability report and puts them into cold storage as spares.
WD (formerly Western Digital) fared much better, but their 3TB drives seem to match up to the results from the 2013 report for Seagate. Both of these lines do run at 7200RPM, so its possible that they just spin out and head on to an early failure sooner than other drives (be it through physical crashes, damaged sectors or something else). As expected, HGST continues on with their incredible performance, posting less than 3% failure rates for all of their drives inside Backblaze’s facility.
However, Backblaze says that 4TB drives, be they standalone or inside external shells, are the best bets for consumers today. Failure rates for 4TB drives from Seagate or HGST are both well under 5% yearly and they do not expect these failure rates to increase dramatically as more drives are added into the pools. Projected failures for the 4TB Seagate drives for 2015 are still under 3.5%, so that’s more than good enough for me or anyone else to recommend them.
WD 4TB drives were excluded from the study because they have so few of them and they do not typically buy up 4TB Red drives because they are more expensive than HGST or Seagate’s offerings. However, with existing failure rates for the 6TB drives already under 5%, it should be perfectly fine to deploy these into a office or enterprise environment.
Overall, it looks like things are still pretty good in the hard drive market when it comes to reliability. Even with the apparently high failure rates of the 1.5TB drives from Seagate, it has to be noted that many of those drives, if not most, are more than four years old and that close on 86% of the drives they have right now will reach 5-6 years of 24/7 operation. That’s quite an achievement for a drive that was never designed to be used in server environments in the first place.