… disk-makers' claims for longevity
I'll bite; which disk makers are these? I bought an OCZ SSD the other day, 6 month warranty. Seagate Barracuda drives - 1 year warranty. It's laughable - it used to be 5 yr warranties by standard.
Cloud backup outfit Backblaze has cobbled together all the data it's gathered from the 25,000 or so disk drives it keeps spinning and drawn some conclusions about just how long you can expect disks to survive in an array. The study's not the best of guides to data centre performance, because Backblaze happily makes do with …
One year on a Seagate Barracuda? That sounds like an OEM drive - one intended to be sold to system builders, who intend to handle their customer's warranty themselves. The consumer warranty on a Seagate Barracuda is 3 years I think, I'm positive it is 2 years minimum.
If you put the serial number into Seagate's warranty website it should tell you what sort of warranty it has.
Yev from Backblaze here -> We are currently trying to gather stats by manufacturer to see if we have enough for a scientific study. If we do, or when we do, we're looking forward to publishing that as well. We have a variety of drives in the data center, so there's a lot of stuff to work through!
I know there's other factors like vibration, impact, and heat, but I'd like to see who has those amazing drives at the other end of the bell curve, the ones that last 100 years. We seem to get a lot of the drives that make it about 1-2 years. Western Digital drives being the worst lately, followed closely by Seagate's "Momentus Thin" series...
Laptop drives in 24x7 and/or array operation are a massive no no in my 15 years of experience. You will only see 12-18months before they kick the bucket, 24 if you are lucky. The Hitachi 24x7 rated ones work better though they have a nasty habit of replacing failed drives with ones that are not 24x7 rated. The only reliable solution I have at the moment is to buy enterprise drives, Seagate Constellation drives seem to work well, but are 15mm deep.
for a meaningful study of hdd lifespan.
I think we would need the date and place of manufacture as well as the make and model to get really interesting numbers.
Two years ago we had a lot of identical drives fail within six months of each other. This represented about 70% of the batch we purchased. The other 30% are still spinning happily. Oddly the failure cluster represented drives that had been in service for five years as well as drives that had been in storage and been installed for a few months.
Drives fail shortly after the warranty expires. Vagabondo's drives still had the warranty expiry timer set to 5 years.
I scrapped a laptop last year when its HD failed after 6 years (not replaceable because the design has changed fundamentally in the interim). The spooky thing was that I had cancelled the extended warranty on 18 September, only to have the HD fail totally on 23 September, just 5 days later. How did it know?
Keep in mind that BlackBlaze uses those drives a heck of a lot more than the typical consumer applications where the drives spend all their time just track following and staying much cooler. Seriously, heat is the enemy of disk drives and for a typical consumer application consumer grade is fine. You want continuous operation you need something with a better thermal profile and those are enterprise drives.
.. that this is a report on consumer grade drives being run in a datacenter operation - i.e. 24x7 at constant (presumably good) temperature and with low to zero accesses occurring. Which doesn't match what happens in a consumer device. e.g. Typically a PC gets turned on, and the cold drive is immediately thrashed senseless by an OS and apps starting, used for some amount of time and then shut down again later. Presumably a /typical/ consumer drive spends most of its time powered off.
With an 11.8 percent AFR in any of the four years, never mind earlier failures, the MTBF cannot be even close to the 2 million hours claimed by vendors. The number would be worse than 0.2 failures in 35,000 hours, which is 150,000 MTBF.
BlackBlaze may be using consumer drives in an enterprise mode, which increases duty cycle, but decreases start-stop. You have to read the small print. Even so, the drives don't seem to meet spec.
I'd believe SSD MTBFs, though. They should meet or exceed 2 million hours without a problem, except for early life failures due to manufacturing.
for once a report that matches my experience both with enterprise drives and commodity units. I have not observed much difference in drives made after 2005 or so. All tend to die after 4 years. Meanwhile, an ancient SCSI doorstop from a CDC of all things still spins up. lights dim and it has not seen active service for a decade. Same fro stuff in ancient AIX boxes that are not as old. Perhaps short life is simply the consequences of increased density and thinner gates in electronics ? All of which brings to mind reliability of offline backup to virtual tape libraries. RAID 6 every where or RAID 10 as a minimum requirement ? Having lost 3 drives in a RAID 5 in quick succession, I still trust tape and mixing batches of drives so a batch fault does not toast customers data. Unfortunately, that is expensive.
Especially with all this nonsense over RoHS (hello tin whiskers), and cutting corners by making in China vs Taiwan or the Phillipeans, even enterprise drives are starting to suck. A certain large cloud provider decided to run their JBOD chassis in such a way as to suck in HOT air and wondered why drives would fail by the bushel. I always spend the extra $20 and buy the SAS models.
Anyone want to help me write an erasure coding module for Linux MD?