Shulou(Shulou.com)06/01 Report--

This article to share with you is about how to predict the failure of the hard disk, Xiaobian think quite practical, so share to everyone to learn, I hope you can read this article after some harvest, not much to say, follow Xiaobian to see it.

We all know that most mainstream vendors currently advertise that they can achieve HDD failure prediction, so that data can be pre-COPY before HDD failure occurs, avoiding the performance degradation and data loss risk caused by RAID reconstruction.

The HDD firmware contains some data provided by hard disk parameters called Self-Monitoring, Analysis, and Reporting Technology (SMART). All array manufacturers say that the S.M.A.R. T parameters predict hard disk failure, but no manufacturer will tell you which parameters and what algorithm to use to make the prediction. In fact, this prediction algorithm is the real core secret, which is very critical to the accuracy of prediction. Hard disk is not bad you predict bad, both for customers and yourself is a loss, because you need to replace the hard disk in advance. Hard drive is going bad you can't predict it, bring risk to customers. These predictions require a lot of use to verify, and constantly improve the prediction algorithm. We see that many new storage vendors also advertise the ability to predict hard disk failure, but it is difficult to play value without a certain accumulation.

Today, let's take a look at the research results of BLAZE, a US cloud storage service provider. BACKZE now has 40,000 hard drives and 100 petabytes of data, so their statistics are somewhat convincing.

BACKZE doesn't know if you still remember, watermelon brother previously shared their statistics on the bad disk rate of various hard disk manufacturers. Simple look, it seems that Seagate's hard disk failure rate is the highest, and later a lot of big V wrote a variety of analysis, Seagate should also do emergency public relations, so it should not have too much impact on Seagate, otherwise Seagate Q3 growth will not be so fast.

The following statistics actually have some value for everyone. The hard disk failure rate in the first year is 5.1%, the second year is 1.4%, and the third year rises sharply to 11.8%. It seems that the failure rate of hard disks of cloud storage service providers is high after 3 years. Of course, enterprise applications shouldn't deteriorate as much due to different business models.

SMART has more than 200 parameters in total. Since there is no standard, many parameters are defined by manufacturers. Therefore, storage manufacturers must cooperate deeply with hard disk manufacturers to mine useful information for themselves.

BACKBLAZE has found that among more than 200 parameters, only the following 5 parameters are strongly correlated with hard disk failure. BACKBLAZE also uses only these five parameters to predict hard drive failure.

SMART 187, for example, indicates the number of unrecoverable read errors (ECC does not recover). If this value is greater than 0, the hard drive will fail. The higher this value, the higher the average annual failure rate.

SMART 12 represents the number of times the hard disk power is switched on. BACKBLAZE found that there was no linear relationship between the number of power switches and the failure rate of the hard disk, so this parameter was not used to predict hard disk failure.

This actually gives a lot of manufacturers who do cold storage a reassurance, because these stores have the function of hibernation, when some cabinets hard disk is not used, you can let the hard disk hibernate, and even turn off the power supply. I was worried that frequent switching of power supply would affect the life of hard disk, but now it seems that it has little to do with it.

Sharing this knowledge today is to tell you that HDD water is actually very deep. All new storage vendors pay a lot of tuition for HDD usage. Therefore, I also give you a few suggestions:

For reliability reasons, major storage companies are required to purchase their own hard drives. Because different hard disk manufacturers have different definitions of SMART values for different firmware. Many users do not understand, why are Seagate hard disk, I can not buy their own? Knowing this knowledge, I believe you will understand the "hidden rules" of this industry.

If possible, choose the manufacturer with the largest shipment volume. Only when the shipment volume is large can we cooperate deeply with the hard disk manufacturer and get more details of SMART data. And the large shipment volume indicates that the hard disk life prediction function of these storage manufacturers has been verified by the market, and its prediction algorithm should be more effective.

Of course, if SDD is used, there will be no such troublesome problem. This is also the advantage of SDD. Life prediction is more accurate.

The above is how to predict the failure of the hard disk, Xiaobian believes that some knowledge points may be seen or used in our daily work. I hope you can learn more from this article. For more details, please follow the industry information channel.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.