Making a more reliable hard drive

[Richard_Cranium72]

MTBF - Mean Time Between Failure. This is the duration of on time that the drive is meant to continue working. (MTBF) is a statistical calculation indicating the mean time between randomly occurring hardware failures. Two parameters are necessary to fully describe how long a piece of hardware will last. The first parameter is MTBF which is a measure of frequency in which random hardware failures will occur. The second parameter is mean operating life which defines how long the hardware will last before an anticipated wearout phenomena will occur. These two parameters combined together give the true projection of the 'real' life of the drive. As an example of how these parameters interrelate, assume your drive has an MTBF of 300,000 hours and an operating life of 5 years. The drive will operate uninterrupted until failure (such as a file server, for example). This is telling you that your drive should be very reliable until wearout occurs since the MTBF greatly exceeds the mean life. However, after 5 years (on the average), expect it to fail due to wearout. In this example, the actual chances of the drive lasting 3 years is 92%, 4 years is 88%, 5 years is 56% and 6 years is 35%.
A read/write head attached to an actuator arm actually floats on a cushion of air, 1-2 micro-inches (one millionth of an inch) above the surface of the platters. A slight nudge, a power surge or a contaminant introduced into the drive may cause the head to touch the platter, resulting in a head crash. The current tolerance drives is 1-2 micro-inches (millionths of an inch). Comparatively, a speck of dust is 4-8 micro-inches and human hair 10 micro-inches. 15 to 20% of the hard drives on campus fail per year?
JE: It's more of a software failure or corruption of the disk file allocation tables, rather than a hardware failure. When I say software, I mean that the machine tries to start up and it can't find the active system folder, or for some reason the active system folder is damaged. That can happen because of an application like Netscape that creates a lot of cache files. Those cache files tend to cause a B-tree corruption problem. The B-tree is part of the disk structure [i.e., part of the software system that keeps track of file locations], and if that gets fouled up, then your hard drive's going to have problems at start-up. http://www.wiu.edu/users/mifdo/hd.html There are two kinds of bad sectors:


A soft bad sector comes from the hard-drive formatting wearing out. You can map it out, or fix it by reformatting the hard drive. Remember, if you reformat the hard drive, you lose all your data.

A hard bad sector cannot be fixed. Data will never be able to be written to that sector of the drive. If you have hard bad sectors, it's a sure sign the hard drive is dying. http://search.zdnet.com/cgi-bin/texis/zdhelp/zdhelp/search.html?Utext=Hard+drives&Uhcat=Hardware&b=tipzone&Utiptype=answer and numerous Hard Drive failure questions here=>> http://computing.net/cgi-bin/AT-search.cgi

What is the most common cause of hard drive failure? Heat! Hard drives keep getting faster and faster. Viurses of course can destroy a HD. I'm not a big fan of leaving the thing running 24/7 either, Have ya ever held a gyroscope with it spinning, the HD tries to duplicate that motion, rough on bearings.. running short on time, gotta run
DrVette

[KillerBug]

I had a great idea for making a hard drive last longer, much longer while riasing costs only slightly, but I need to know, when a hard disk dies, what happens? I heard my old 200mb start clicking and I have heard the same problems from others on the forum that have lost newer drives, but I also lost a 500 without any clicking, just stopped working all together, I thought to open it up and it looked fine, no loose peices or anything, the heads were even intact. So What makes a drive truely die? (not just get errors)

[grunt]

Spindle motors burn up, logic boards die, if the read/write heads come into contact with the platters that will kill it too. Heads can also wear out and lose their ability to read - so they just try to seek, fail, click back over and try again.

I took a clicking Samsung apart once and fired it up. It was interesting to watch the heads try to seek, then slam into the spindle http://sysopt.earthweb.com/forum/smile.gif

[KillerBug]

So to make a truely reliable hard drive, you would have to:

1.) Have backup chips
2.) Have active cooling
3.) Have a backup motor
4.) Not alow the heads past a certain point, like with a bar that stops the arm.
5.) Have backup heads.

That about right?

[qball]

Solid State.

Putting rust on sticky tape is getting pretty friggin old.

Wait, we can all use paper!