SMART: A Failure May Be Imminent

Written by Marc Roberge | 03 August 2011

When booting your computer, you may have noticed the word SMART (or rather the acronym S.M.A.R.T.) pop up on the P.O.S.T. screen (white text on a black screen) and paid little mind to what it is or what it does. That is of course until, one day, your computer doesn’t happily chug along and boot into your Windows, MAC OS or Linux install, but rather, halts on the post screen and displays a warning predicting imminent hard drive failure and prompting you to perform a backup ASAP. Instantly, you transform into the geek equivalent of Jack Bauer as you race against the clock to get your data backups done and your hard drive replaced before your world comes crashing down around you.

But wait! Do you even know what S.M.A.R.T. is? How does it “know” your hard drive is about to fail? And how accurate is this prediction? Is it some highly subjective Nostradamus type of prophecy?

S.M.A.R.T. stands for self-monitoring, analysis and reporting technology and it was developed in the mid 1990s by Compaq (and its partners Seagate, Quantum, Connor) as well as IBM and Western Digital as a system of metrics for predicting hard drive failure. The objective was to be able to both collect data that would help in designing better, more reliable drives AND provide end users advanced warning of impending failure.

On the other hand, there are a few things you should know. First, there is no hard and fast set of SMART standards across drives and manufacturers. Features and metrics found on one drive may not be included on others. Additionally, depending on the type of interface used, some SMART enabled motherboards may not be able to communicate correctly with hard drives (ie via USB or Firewire).

Among the diagnostic data your hard drive may be logging, there may be somewhere in the area of 50 different metrics being used to monitor the health of your hard drive such as Seek and Read/Write Error Rates, Throughput, Temperature and or Shock. For a more comprehensive list, read SMART on Wikipedia. One interesting thing I came across was a study conducted by Google Labs who performed independent testing on 100,000 hard drives, the largest such study ever conducted. There’s a lot of good information in their report that I won’t get into, suffice to say that one of their determinations was that SMART only managed to predict about 56% of drive failures and that, on the flip side, while all drives that reported SMART errors eventually failed, there was no predictable timeline.

So? What does that mean?

Well, common sense would tell you that hard drives, being the mechanical and electronic wonders they are, will eventually fail. That’s a given. Every “expert” out there will tell you that, if your drive is reporting SMART errors, you should replace it ASAP and toss the old one out. The issue is, however, how much stock should you, the consumer, put in SMART error reporting in the first place? Let’s look at it from a consumer stand point! On the one hand, you have hard drive manufacturers that design, produce and sell a technology that has a limited, albeit unpredictable, lifespan. The more drives they sell, the more money they make. Of course, make too many crappy drives and you’re not making ANY money. So they develop a signaling and data logging technology called SMART, not to prevent but, to predict drive failure. The catch is…how accurate or useful is the data being logged?

SMART is a hardware/software data logging system and, as such, is prone to it’s own errors or inconsistencies. And what kind of independent testing has been performed to ascertain that the particular metrics being used significantly contribute to hardware or mechanical failure? Would you implicitly trust ANY manufacturer to not build in some type of obsolescence or disposability? The reason I say this is that I’ve had many, and by many I would estimate over 20, hard drives over the past 10 years or so that functioned without fail several years after warning me of imminent failure. Naturally, when a drive throws a SMART error, I do the SMART thing and replace it but I never throw a drive out until it actually fails. This means that I’ve put these drives of impending doom into external enclosures, internal secondary and tertiary storage in my computers, even built projects around them (test environments, media centers, laptop robots) and I would say that only about 30% (an estimate) failed within a few weeks or months following the first SMART error warning.

That means that, at least in my own personal experience, 7 out of 10 drives that SMART predicts will fail and urges me to replace actually function reliably and DO NOT need to be replaced. Even if the number were closer to 50%? It still begs the question, how reliable is SMART? I suspect that for the average consumer, this discussion is a moot point – why would anyone be bothered when replacing a drive is a matter of a few bucks and a few hours? For me, it’s a matter of ownership. If you are going to implement a system such as SMART, as an end consumer, I should have the ability to disable, clear or reset this system. As it stands, I don’t know the nature or circumstance of specific errors and I cannot clear the data to diagnose issues for myself.

SMART: A Failure May Be Imminent!!!

So what can you do should your drive start throwing up SMART errors? The short answer is…not much! You can choose to ignore it, cross your fingers and press F1 (F2?) every time you boot your computer or replace the drive. As the “experts” (trolls) will tell you anytime you ask about it on any support forum…SMART is telling you your drive is going to fail…duh! What no one seems to want to discuss is how SMART isn’t always so SMART and that, while there isn’t a lot you can do, there are some things you can explore to see if you can tempt fate and extend the life of your dying drive.

Some, not all, BIOS environments allow you to simply disable SMART to allow you to bypass the nagging halt screen on boot up should you ever receive a warning. In other instances, with standard ATA drives, you may be able to disable or modify SMART logs directly on the drive using the Hitachi Drive Feature Tool. Barring those 2 options, there are utilities available that will allow you to see the type an nature of the error or errors and decide for yourself whether they are significant enough to prompt you to toss the drive (check out HDDScan).

Hope this helps!