What hard drive manufacturers aren’t telling you…

Douglas Smith, president and co-founder of Network InstrumentsDouglas Smith, president and co-founder of Network Instruments, blogs about what the hard-drive manufacturers aren’t telling you with regard to hard drive failure rates.

There seems to be an incredible disconnect between published hard drive failure rates (MTBF, etc.) and what end users can expect in the field. I can only speculate on how this whole thing got so far from reality, but I suspect that one vendor began fudging the numbers and the next company had to do the same to remain competitive. To be fair, they all have their fine print regarding how the testing is done, but unlike automobile gas mileage ratings, people seem to take hard drive failure ratings seriously. The Daily Tech wrote a great summary article of a Carnegie Mellon study that was done on the subject. I think reading the summary article provides enough information (the study is a bit lengthy).

In a nutshell, the study states that drive failure rates are between 2-13% per year. We have seen other studies that state average failure rates are around 5% per year. Keep in mind that these rates are for average use (studies include data centers and user systems). Hard drive manufacturers would have you believe the average rate of failure (MTBF) is over 1.5 million hours. This equates to roughly over 171 years of constant service before problems. A 5% failure rate translates to 100% failure in 20 years, a far cry from 171 years…

Now let’s turn to the environment in which retrospective network analysis (RNA) devices capture data at network rate, 24x7x365. For drives in our product (GigaStor) and for any other RNA type device, this environment is the absolute worst-case scenario. While we have not seen drive failures as high as 13%, it should be expected that an RNA device’s hard drives will fail on the higher end of the scale.

One last note: the study indicated that the highest failure rates are in the first year and after the fifth year. Translated, 2-13% is an average over the life of a drive – in the first year and past the fifth year drive failures will be much higher.

We take responsibility for all product failures, regardless of sub-manufacturers claims. I suspect all RNA device manufacturers do the same. But it is clear that companies like Maxtor, Seagate (who now owns Maxtor), etc., are doing us all a disservice by making such ridiculous reliability claims. I guess for now it is up to us RNA device vendors to clean up the mess the hard drive manufactures have left us with.

Digg What Hard Drive Manufacturersdigg Bookmark Google Cheat Sheet on del.icio.usdel.icio.us Add Hard Drive Manufacturers to your Technorati Favoritestechnorati

2 thoughts on “What hard drive manufacturers aren’t telling you…

  1. I thought that failure rates are generally assumed to be Poisson distributions, meaning that the failure rate over time is constant, regardless of how many years the part has been operation.

    In that case, a 5% failure rate (one year) would imply a 96.5% failure rate over 20 years (that’s 1-(0.95^20)).

    Now, if you want to take exception to the practice of assuming failure rates follow a Poisson distribution, that’s one thing. But this just looks like bad math.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s