Barracuda XT, Barracuda, Barracuda Green, DiamondMax and other desktop SATA and IDE drives

[VeganFanatic]

I noticed the the other guys red series disks are being flogged with 1,000,000 hour MTBF which is much higher than I have expeienced.

One reason my values are lower is that every time I get a new hard disk the old one ends up as a spare in case of a failure.

So far, knock on wood, there have been more more problems with sick disks following the 2x 500 GB disks that croaked. I have more than doubled the storage since those died.

I will be using more disks on my server so it can be used for backups etc. I am thinking of a SSD for the boot disk to speed up the web server a bit,

Given the number of operating hours in a year, 1,000,000 is a lot of years, I am happy to have disks rack up 100,000 power on hours as by that time that level is reached the disk is obsolete.

I have noticed that no disks larger than 3 TB are available, suggesting that the current perpenedicular design has hit the wall.

Lots of ideas out there, so my thinking is using lots of 3 TB disks for storage

[fzabkar]

MTBF (Mean Time Between Failure) has a counterintuitive meaning.

AIUI, an MTBF of 1,000,000 hours means that in a population of 10,000 drives, say, it is expected that there will be 1 failure every 100 hours. For 1000 drives, the failure rate would be 1 failure every 1000 hours. In other words, if a large batch of drives runs for a combined total of 1 million hours, then there will be 1 failure in the entire batch, on average.

My understanding is that MTBF falls as the population ages. For example, it could be that after 1 year the MTBF reduces to 500,000 hours.

See http://www.adrc.com/ckr/mbtf_hdd.html

[VeganFanatic]

I have a somewhat smaller population of disks, and I only procure a disk or 2 a year on average.

My ST2000DL003 is the last disk I bought, so far so good.

Probably get a 3TB disk to add to the existing 2 TB disk, I put big disks on the server for backups etc

AFR is another metric some vendors use. 1 million inplies a very low AFR

[fzabkar]

Here is an excerpt from Seagate's documentation:

Barracuda ES Serial ATA Product Manual, Rev. B:
http://www.seagate.com/staticfiles/support/disc/manuals/enterprise/Barracuda%20ES/SATA/100424667b.pd...

2.12.1 Annualized Failure Rate (AFR) and Mean Time Between Failures (MTBF)

"The product shall achieve an Annualized Failure Rate (AFR) of 0.73% (MTBF of 1.2 million hours) when operated in an environment that ensures the HDA case temperatures do not exceed 40°C. Operation at case temperatures outside the specifications in Section 2.9 may increase the product AFR (decrease MTBF). AFR and MTBF are population statistics that are not relevant to individual units."

Crunching the above numbers, we get ...

MTBF = 8760 hours per year / AFR = 8760 / 0.0073 = 1,200,000 hours

Using the subject MTBF, we get ...

AFR = 8760 / MTBF = 8760 / 1,000,000 = 0.88%

[VeganFanatic]

I am aware of statistics, I have taken all of the courses and lots of advanced math.

I am more interested in the hard error rate, the ES drives are better at 10^15 but I wonder if that could be bumped up with modern high capacity disks

Right now for bulk storage, these low cost green series disks seem adequate. The disk is not very havily loaded most of the time, depending on the web server load.

it hosts download content for my chess site

I am more cynical that a disk can last 100,000 power on hours or not

[fzabkar]

Sorry if I come across as a pedant. I often pitch my responses to a wider audience, many of whom may not be familiar with the terminology.