SSD drives are super unreliable – or are they?

graphical representation of a broken SSD

Posted on Wednesday, June 8, 2016

Author: CDS

A recent study published by IT Pro Portal has been gaining a large amount of attention, due to the headline claim that SSD drives are far less reliable than traditional spinning disks.

At a time when many enterprises are rolling out SSD replacement and upgrade programs, such statistics are particularly worrying.

The survey claims that 92 per cent of respondents are using SSD technology, 38 per cent have experienced a drive failure, and 23 per cent of those failures resulted in data loss.

But are these stats as terrifying as the headlines suggest?

Who are these users?

The first thing to note is that the study makes no distinction between consumer and business users – which raises the question, who were the 2000 people questioned?

Consumer-grade drives tend to be inferior to their enterprise-grade counterparts for instance – so the stats could be heavily skewed by questioning the “wrong” people.

How has the SSD technology been implemented?

With no moving parts, SSD drives are impervious to the physical failings of their spinning disk counterparts. SSD drives are affected by wear inside the physical flash cells however which does shorten the lifespan of the drive.

Most enterprise grade drives are built to reduce these wear failures if they are configured correctly. SSD drive failure can be accelerated by putting flash drives behind parity RAID controllers. When the controller detects a PE error caused by wear, it immediately begins a drive rebuild, rather than allocating a spare block from the overprovisioned PE pool. Worse still, parity RAID rebuilds dramatically shorten the lifespan of the drive.

Which means that some SSD deployments are actually designed from the outset to lose data. And that proper configuration of SSD RAID arrays will help to reduce drive unit failures.

But there are lessons to be learned

So it would seem that the Kroll OnTrack survey raises more questions than it answers – a classic sign that hysteria has overtaken reason, and allowing the CTO to sleep a little easier.

That said, SSD failure and data loss is still a genuine concern – even if the failure rates in the enterprise are much lower than these statistics suggest. It could be that the solution is to use disk to disk backups, with low cost, high capacity spinning disks being deployed as an archive, sitting behind your SSD production systems.

As SSD deployment programs ramp up, there is a credible business case for redeploying redundant spinning disk arrays for exactly this purpose. Even post-warranty hardware (backed by a suitable third party maintenance contract) could play a part in protecting against data loss and improving access to archive or backup data.

To learn more about preventing data loss from your SSD arrays by redeploying existing storage assets, book your place on our free webinar on June 9th.