How many read/writes hard disk's endure

  • Thread starter Thread starter avsrk67
  • Start date Start date
A

avsrk67

Hello Folks ,
How many read /writes does a typical hard disk (SATA or IDE) would
endure .
This is not mentioned in any spec from the manufacturers .
Thanks
Subra
 
Hello Folks ,
How many read /writes does a typical hard disk (SATA or IDE) would
endure .
This is not mentioned in any spec from the manufacturers .
Click to expand...

Can you be kind enough to explain to me what you mean by "endure" before I
can make some wild guess?
Thanks
Click to expand...

Thanks for the explaination too
 
Hello Folks ,
How many read /writes does a typical hard disk (SATA or IDE) would
endure .
This is not mentioned in any spec from the manufacturers .
Thanks
Subra
Click to expand...

I suspect that "this is not mentioned in any spec" because it doesn't
apply. A drive that is working well will fail not because of how many reads
and writes it has done but for mechanical failure or electrical breakdown,
neither of which have anything to do with reads and writes.

Example: I connect a drive to power and spin it up and never do a single
read or write but simply rack the heads from the innermost to the outermost
tracks as quickly as they can move. Eventually the drive will fail from
bearing or motor or positioning or electronic breakdown. Reads and writes
never enter into it but operation time does. That is why you see Mean Time
Before Failure (MTBF) quoted in specs. And even then this is only a
statistical guess. A drive with a quoted MTBF of three years could also
fail in three months or three decades. Neither of these extremes is likely
under the expected conditions but they are certainly possible.
 
Hello ,
I mentioned read/writes because that is what would make the head
move .
I have a system that would be writing 50 k of data every second .
Any guess how long would that drive last before it gets corrupted .
thanks
Subra
 
I don't think it's much or any difference between these days and how hard
drive worked back in 70's when I worked for company making hard drive. And
we always had dozen hard drives without cover running (testing I believe)
24/7, and I can tell ya that there is no answer to your question.

The disc spins at hi-speed, when the head flies above the surface. The
head crash is pretty rare and often cause by the arm (and could be the
head), and the failure often cause by coating formular (as well as
processing like coating, buffing etc.) and that's why we notice some brand
(manufacture) is better than other.
 
Hello ,
I mentioned read/writes because that is what would make the head
move .
I have a system that would be writing 50 k of data every second .
Any guess how long would that drive last before it gets corrupted .
thanks
Subra
Click to expand...

For some real world feedback you might try writing an email directly to a
couple of the system operators for newsservers. If anyone has seen tested
drives to destruction with read/writes their systems would.

As for worrying about the data being corrupted, if it were me I'd put
together a system utilizing a Raid 5 system. That way should one drive fail
you could replace it and not loose any data. Then if worried you could
cycle replacements for the other drives one-at-a-time and let the system
rebuild the data structure while still using the array. The right raid
system monitor would email or page you when the array needs a little
personal help, as in a drive failure. The other trick is to keep the drives
as cool as possible, the tend to fail sooner as the temperature goes up.
 
Hello ,
If we are writing once every second in 10 days we are approx close to
a million writes .
In 100 days (3 months ) close to 10 million writes . Is that close to
failure point of disk .
Of course i am evaluating RAID 5 .
Thanks
subra
 
Hello ,
If we are writing once every second in 10 days we are approx close to
a million writes .
In 100 days (3 months ) close to 10 million writes . Is that close to
failure point of disk .
Of course i am evaluating RAID 5 .
Thanks
subra
Click to expand...

I have never seen technical information that addresses the life span of
the flex cable, connecting the heads on the actuator, to the controller.
Presumably there is some limit to the number of times it can be flexed.
Since there are many disks used in database applications, and they
do a couple hundred IOP per second, the drives would appear to be
able to handle that kind of abuse.

http://www.dansdata.com/images/magnets/voicecoil440.jpg

It is too bad they couldn't do a study like this, and do a
post mortem on each failed drive.

http://www.siliconsystems.com/technology/pdfs/Hard_Drive_Failure_Studies.pdf

Paul
 
Phil B said:
Google released an interesting report about disk reliability recently.
Google for 'google disk reliability'

One of their counter-intuitive conclusions was that disk temperature was not
an issue. I suspect that only applies to
24/7 running and that aggressive temperature cycling in a poorly cooled
domestic situation might be very different.

I would think that other factors may be more important than raw drive
reliability.
You will have to factor in software and power supply reliability. Software
crashes
and power fails can leave corrupted file systems through incompete closure
of files, even though the underlying hardware will be perfectly intact.
Human error is also a significant factor - an error on partitioning or a
clumsy delete. I had a situation once where every disk
in the computer room (all four of those huge 30MB drives!!) had to be
re-formatted and re-written from scratch just because someone typed a wrong
date!

Phil
Click to expand...

My main system has always run 24/7, and I have been doing this for
decades. Years ago, when I was running the BBS I always had a fan running
to keep it cool, now DVD authoring I have installed 2 Jet Fans (powerful
fans with fan controller) and 2 extra regular case fan (I don't remember if
any come with the case besides the power supply). That's how I keep my
system cooler (it's still hot especially in summer month)

The only problem I have with heat is the Heatsink Paste which will turn
into ASH after few years of 24/7 services.
 
kony said:
Silicone oil based pastes are the worst for that, avoid them
when possible.
Click to expand...

I often use the silicone comes with the CPU. I believe I have bought a
tube with good feedbacks (several years ago) but haven't had the chance to
use it. Or by the time I built another system, I either forgot or can't
remember where I put it.
 
Back
Top