[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: OT: Avoiding harddisk failure due to mobility


On Mon, 2003-04-28 at 14:30, Rahul Kalaskar wrote:
> Hi All,
> I need some advice on finding a solution to this problem . I have a hard
> disk mounted on a mobile robot. I am using ext3 fs. The hard disk is
> subjected to several bumps as the robot travels. I am experiencing
> frequent hard disk failure ( about once a month). I have thought about 
> 2 different solutions to this problem. I would appreciate any tips on
> evaluating them

	Have you considered laptop hardrives? I believe that they could be
built with a little more ruggedness then standard desktop drives, which
I believe that you are using. They might last longer.

> 1) Use Compact Flash cards: This is a an easy solution. However, I need
> about 1GB of space. 1GB CF cards cost about $300. There is the issue of
> CF wear. The cost is the major con of this approach.

	CF cards sound like an excellent solution, but your are very correct
about the cost...

	Another thing you could look at are the Solid State hard drives. I
understand that those have dropped in price a bit over the past few
years. They might be more affordable for your product. Another thing to
look at, is to see if you could locate some technical specs on one of
those "new" Sony Robots, the robots that look like "Twiggy" from the old
US Television Series Buck Rogers (circa 1980's). The guts of one of
those robots could give you an excellent idea of which direction would
be best to head towards.

> 2) Using Ram disk. 
> * Create 1 or 2 ram disks on boot.
> * Mount the stuff that gets written ( /var/log, /tmp, some /etc related
> to ssh) on the ram disk. )
> * Put the hard drive to sleep, while the robot is in motion.
> * Sync the logs from the ram disk to the hard drive, when the robot is
> stationary.

> Approach 2 assumes that, if the hard drive is sleeping while in motion,
> it will not be damaged. Is this true ??
	I believe that you will still have issues with the hard disk failing,
even if it is asleep. If I recall correctly, a good deal of hard drives
can only handle a few "G's" of force, somewhere around 5 or 6, when not
powered up. You can actually produce that kind of damaging "G" by
sitting a hard drive up on it's end and then simply tapping it enough
for it to fall over on its own.

	That is a HUGE reason why Seagate started covering their hard drives in
a rubber "skin" it isn't to practice "safe computing" so much as to
practice "safe building" of computers. They weighed the costs of taking
back returns of trashed hard drives from people tipping them, while
building PCs and the costs of tossing on a little rubber suit for their
hard drives. They found it less expensive to go with the suits, even
with the additional thermal wear that the rubber suit will cause, due to
it's insulating properties.

> Does approach 2 sound reasonable? Any other tips/suggestions/criticism  

	I can only suggest building some kind of "shock cage" or other shock
absorbing "thingy" to hold the hard drive in space. Sort of like the
older "Use your portable CD-Player" in your car devices. Basically, you
would drill a hole in the floorboard of your car to mount a rod with a
special shock absorber on it. The CD-Player sat on top of that and it
was supposed to keep the CD from skipping. I don't know how well they

	Look up the "Egg Drop" experiment. A number of the better high schools
and more 1st year engineering students have to perform this experiment.
Basically, they are given an egg and a box size. Then they have to build
something within the box to keep the egg from breaking, when the egg is
dropped from roughly 10 to 15 feet in the air to a, typically, cement

	Those are my only suggestions.

> Thanks!
> Rahul

	You are welcome,
	Robert Adkins
	IT Manager/Buyer
	Impel Industries, Inc.

> _______________________________________________
> Ext3-users mailing list
> Ext3-users redhat com
> https://listman.redhat.com/mailman/listinfo/ext3-users

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]