Feeds

Server, server in the rack, when's my disk drive going to crack?

Backblaze's 25,000-drive study scries the future of your storage

High performance access to file storage

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 consumer-grade drives. As even those drives routinely offer mean time between failure (MTBF) in the hundreds of thousands of hours – decades of operation – or the storage industry's preferred longevity metric of annualised failure rates (AFR) of under one per cent per year, the study tests those claims as well as any other. It's also rather more recent than the 2007 studies from Google (PDF) or Carnegie Mellon University.

Backblaze's study finds that both AFR and MTBF are bunk. The document finds that disks follow the predicted “bathtub” curve of failure: lots of early failures due to manufacturing errors, a slow decline in failure rates to a shallow bottom and then a steep increase in failure rates as drives age.

Disk failure rates from Backblaze's disk longevity study

Backblaze's disk longevity study shows something pretty close to the 'bathtub' curve one would expect

The study then looked at when drives fail and found a drive that survives the 5.1 per cent AFR of its first 18 months under load will then only fail 1.4 per cent of the time in the next year and half. After that, things get nasty: in year three a surviving disk has an 11.8 per cent AFR. That still leaves over 80 per cent of drives alive and whirring after four years, a decent outcome.

The study also predicts accelerated failure rates in years four and five, guesstimating things will get very, very bad in years four and five.

Backblaze promises to compare consumer-grade and enterprise-grade drives in a future study, which will be interesting if it reveals the premium paid for the latter makes little difference to longevity. Whatever the outcome of that study, this one shows that disk-makers' claims for longevity need to be taken with a decent pinch of salt. ®

High performance access to file storage

More from The Register

next story
Seagate brings out 6TB HDD, did not need NO STEENKIN' SHINGLES
Or helium filling either, according to reports
European Court of Justice rips up Data Retention Directive
Rules 'interfering' measure to be 'invalid'
Dropbox defends fantastically badly timed Condoleezza Rice appointment
'Nothing is going to change with Dr. Rice's appointment,' file sharer promises
Cisco reps flog Whiptail's Invicta arrays against EMC and Pure
Storage reseller report reveals who's selling what
Bored with trading oil and gold? Why not flog some CLOUD servers?
Chicago Mercantile Exchange plans cloud spot exchange
Just what could be inside Dropbox's new 'Home For Life'?
Biz apps, messaging, photos, email, more storage – sorry, did you think there would be cake?
IT bods: How long does it take YOU to train up on new tech?
I'll leave my arrays to do the hard work, if you don't mind
prev story

Whitepapers

Securing web applications made simple and scalable
In this whitepaper learn how automated security testing can provide a simple and scalable way to protect your web applications.
Five 3D headsets to be won!
We were so impressed by the Durovis Dive headset we’ve asked the company to give some away to Reg readers.
HP ArcSight ESM solution helps Finansbank
Based on their experience using HP ArcSight Enterprise Security Manager for IT security operations, Finansbank moved to HP ArcSight ESM for fraud management.
The benefits of software based PBX
Why you should break free from your proprietary PBX and how to leverage your existing server hardware.
Mobile application security study
Download this report to see the alarming realities regarding the sheer number of applications vulnerable to attack, as well as the most common and easily addressable vulnerability errors.