MBR and hardware wear and tear
 

I have to agree with leosubhadeep, go with grub. Like he said, Lilo writes to the MBR every time you modify the boot options. One thing about this (besides what leosubhadeep said) is it bothers me having the MBR written over and over again. I know it's a HDD and that shouldn't be a problem but I'm paranoid about such things. The MBR is one very special physical section of the HDD and it's my opinion it should be left alone as much as possible. Grub doesn't mess with the MBR after changing a boot option, it modifys a file.

//These posts were split off from http://www.linuxquestions.org/questi...do-4175427429/ and moved to the Hardware forum.

The MBR is a sector on the HDD like any other sector. If it gets bad it will automatically and transparent to the system be replaced with a spare sector by the firmware of the HDD. Nothing special here.
Besides that I never have heard of a worn out MBR because of too many writes. How often do you change your kernel or your boot configuration?

OK, I have made a little research on the net and found not one case of a worn out MBR.
And I am not surprised by that. As I stated in my previous post, to the disk the MBR is nothing special, it is just a normal sector that can be replaced by spare sectors if it gets bad. So basically, even if something like a worn out sector may happen, the user will be unable to boot the system, but once the disk's firmware replaces it with a spare sector and the MBR is rewritten it will work again as before.
I also wasn't able to find information how many writes a single sector can handle, all information I found pointed out that it is much more likely to have a mechanical failure before wearing out the magnetic material (if this is even possible, I could also not find any information on that).

It would really be nice if anyone with information on that topic could post some links here.

I really appreciate the work you have put into this. Very, very interesting. Since I started using and working with computers in 1999 I have been more interested in the mechanical than the software. Early into my experience with different computer configurations I became paranoid about the MBR because of doomsday stories about how a person's PC mysteriously became incapable of booting until they did something like which was of course MSDOS or Windows. In 2001 I changed to Linux but the fears lingered. They have subsided somewhat but the spider webs are still in the corners, I guess.
Thank you for sharing this information. It is comforting to know losing a sector in the MBR does not render a modern HDD unusable.
About the platters and the magnetic material, it makes sense it could go a very, very long time being written over thousands of times. And the other hardware - the servo, servo arm and read/write heads I have dissassembled and examined look like they could last 50 years easily. The servo and heads are mounted with twin sets of extremely high quality ball bearings. All this in a completely sealed unit. From this hands on research I have determined between 85% and 95% of HDD failures are probably due to the circuit board componentry.

From my experience (private and at the job), many disks die from faulty electronics. But I wouldn't think (damn that I not have kept any numbers on that) that the percentage is as high as 85% or even 95%. I would think that at least about 30% are mechanical failures and maybe 10% are failing due to bad sectors (although that also can be counted to the mechanical failures, for example when caused by a headcrash due to vibration).

But I really find it surprising (and annoying) that I was not able to find even theoretical numbers on the maximum number of writes (or would magnetical change be a better term?) one harddisk sector can take. The only numbers I could find were about SSDs, with a range from about 10,000 writes per cell on cheap MLC drives to more than 100,000 writes on the better SLC drives. But even in the case of a cheap SSD without any form of wear leveling/sector replace I would think that having to overwrite the MBR 10,000 times to get it damaged is more than enough to assure that even in case of using lilo with many configuration changes/kernel updates you will not damage the disk in the ususal lifetime.

Yes I too can see now from what I learned from you that MBR overwrites are not a real problem even with the low quality SSDs. But now I'm obsessed with why there is so little information on how many writes a hard disc can do. I remember once reading in a forum the question "how many times can a DVD-RW disc be written?", to which the expert's reply was "That's a good question", and that was that. Frustrating. I wonder why no extensive research on this has been done? Could it be because since the major culprits are things other than the platters and heads (disregarding things like dropping the HDD on a concrete floor) nobody has ever seen a marketable reason to test these? I mean if you are a HDD manufacturer and you brag about how many writes your platters/heads can do will the consumer not fire back with "yeah, but what about the electronics?"?

Read this:
http://serverfault.com/questions/285...early-failiure
Particularly the answer provided by "Chopper3". On this site - serverfault.com - he has a reputation in the top 0.04 percentile.

If you want stats, look up HDD vs SSD comparisons,... You'll get plenty of figures...

Have you any bookmarks on the subject you've found highly pertinent you'd be willing to share? Thank you.

Even if there is not really a practical way to test this, may be because the electronics or mechanics will always die first the manufacturer's must have some data on theoretical values regarding this topic.
May it be possible that this information is am industrial secret?