Slashdot Mirror
What To Do With a Hundred Hard Drives?
Makoto916 writes "In five years with my current employer as the IT administrator, I've amassed a sizable cabinet of discarded hard drives; just shy of 100, in fact. All of the drives range in size from 20GB up to 300GB. They've all been stored in anti-stat bags, and spot checks of even the oldest ones show that most of them still work. Individually, they're mostly useless for our line of work, which is digital video production. However, the collective storage potential is quite significant. They are of varying size and speed, but the one commonality is they're all IDE. What is the best way to approach connecting all of these devices and realizing their storage potential? On a budget, of course. Now, I'd never use such an array for critical data storage, but it certainly would be useful as a massive backup array to our existing SAN that does store critical data. I have several spare and functioning PCs, but not nearly enough to utilize their internal IDE controllers; even with multiple add-in controllers, it still wouldn't be enough. Not to mention the nightmare of managing a bunch of independent PCs. I've looked into ATA Over Ethernet and there's a lot of potential there, but current 15 to 20 bay AoE cabinets are expensive, and single device enclosures are so rare that they're also expensive. Are there any hardware hackers out there who have crafted their own home-brew AoE systems? Could they scale to 100 drives? Is there a better way?"
e Bay.
I doubt its worth using a bunch of old smaller drives.
between the power requirements and all the extra hardware needed to run them i would just sell them all on ebay and take the $ to buy a couple of huge drives, mirror and do iscsi with them.
---- Booth was a patriot ----
spin around in a circle and see who can throw them the greatest distance
Contrary to popular belief, Unix is user friendly. It just happens to be particular about who it makes friends with.
Granted, you have a few less than others, but it's worth giving a shot
to other employees, give the proceeds to Charity.
There really just a waste of company space and time.
The Kruger Dunning explains most post on
Would be a super generous donation, but if you honestly don't have a practical idea, perhaps donate to your local Free Geek chapter? Good drives at that size could help in the fight for bringing technology to those who couldn't afford it otherwise.
http://mediagoblin.org/
Ebay and use the revenue to buy a few very large size drives. Running a ton of tiny drives on standby all the time just makes no sense from both a power and heat standpoint.
http://www.thementoringctr.org/index.cfm?fuseaction=pages.Digital&
Im sure you could donate the hard drives to them and get a tax writeoff...or
find something similar in your community
With almost a hundred hard drives, the gold leaf discs inside them must really add up in weight. What's gold trading at now? $850 or something per ounce.
Admit it. You post strawman arguments as AC so you get modded Insightful for refuting them, rather than Troll
freenas + old motherboard + all pci slots full of cheap IDE cards.
works great, dont bother with IDE drive size versus Motherboard/Bios as freenas does not use the bios.
I have made a couple of 2TB arrays from less than a couple hundred bucks and a bunch of free 250gb hard drives.
You can do a software raid5 which gives you some peace of mind.
Do not look at laser with remaining good eye.
I can't imagine the storage is worth the time to set up something that can use them all given new 500GB drives They are probably most useful in cheap USB to IDE enclosures as additional external storage - nice for convenient system backups, offsites, and extra storage.
Destroy them. If they stored what you describe, you do not want proprietary information leaking out. Especially, if you are the one that is in charge of "doing something with said HD's". Safer to destroy them.
Of course, all slashdotters would say either build an array or donate. In reality, the company should keep the biggest for desktop usage and shred the rest.
Safer for you and the company in terms of liability.
Pry them open, remove those awesomely strong magnets, and stick them all over some douchebag's Hummer.
A hundred old hard drives stood up on a wall!
A hundred old drives on a wall!
BANG!
Ninety-nine old hard drives....
I got a $20 enclosure with 17 drive bays in it, and a 300W power supply. I've got a dozen SATA drives, each drawing under 10W, and 5 EIDE, drawing under 20W each.
At first I just got a dozen SATA/EIDE USB slaves for $10 each, and plugged them all into a USB hub, with just the single USB cable stretching out of the case over to another full PC's USB socket. But that is so slow, especially when copying big music or video files between drives (and through the single USB cable to the CPU and back). Playing multiple media files to different terminals in my house is too much bandwidth for the single USB, too. Running 4 USB from the big enclosure to the 4 sockets in the server PC isn't much better, because it all goes through the same CPU and PCI bus.
So I got 3 Sabrent SBT-SRD4 4xSATA controller PCI cards, because they were $25 each. But when I tried to boot them in a few different motherboards (pre-HP Compaq P3/1.2GHz, IBM P4/3.2GHz), none of them got past the POST to even start booting the OS. I want to use them with Linux, but with the failure to even boot I'm not hopeful about driver support, either.
I bought them from CompUSA (still alive, online only), which hasn't replied to (email only - no phone available) tech support requests. Nor has Sabrent itself. I'm not hopeful that they'll refund my money, since everything else about this transaction has sucked.
So what I want to know is what cheap motherboard (no need for graphics or anything else other than at least 3 PCI slots and 100Mb-1Gb ethernet) will work with these SATA cards? If they're really duds, what is the cheapest way to get 12 SATA drives controlled, even if they're not that fast, over to 100Mb/Gb ethernet? Either SATA cards + motherboard, or even a fat mobo with a dozen SATA ports. I'd even settle for just 4-8 SATA ports to get started. I'm talking under $200 if possible.
Ideas? If it works, then 8-9 of them should support the 100 HDs the original question was asking about.
--
make install -not war
I think this is how Google started. Throw in some other random hardware and wait for the VC to come rolling in!
Pull them apart, and use the magnets to make a magnetic rail gun. Or some other fun game. There has to be a lot of fun (and destruction) in 200 ceramic magnets.
But you could make a hard disk generator I've seen several designs and some are better than others, but there isn't a great way to string out hundreds of IDE drives without a cluster and multiple processors. After weeding out a number of the large drives for storage, it may be a fun project to mess around with.
Well, back to rejecting software patent applications.
You could build a climbing suit for climbing steel, build a generator,....
Engineering is the art of compromise.
I would call your local data recovery service as they sometimes are interested in buying old drives of no particular size to use the controller cards on them.
Apparently, a lot of failed hard drives are not bad because of their physical platters, but because of the drive logic. These places need old drives for replacement controllers that you probably can't buy from the manufacturer.
ft
Pick the largest and buy up as many usb or firewire interfaces and drop them in a tower case with a psu for the HD's (get bus powered usb/firewire interfaces) and have a decent sized external array...
or use the larger ones as customer throwaways - when the video needs to go to the customer and its really big - ship them a cheap usb/firewire enclosure with a disc in it loaded with their video - if it doesn't come back then you've got more to spare....
Life is but a Beta test...
power_to_run_100_hard_drives = 100 * power_to_run_1_hard_drive
Just keep in mind these are *STRONG* magnets. When you take it apart the magnets may smash into each other. This could send particles flying away in a direction that, according to Murphy, is where your eyes are. I know this by experience, lucky for me I wear glasses. And if some of your flesh is between the magnets, it's painful.
I'm actually thinking that it's a waste of effort. If they average, say, 100Gb each, then 100 drives means 10TB. 10TB these days is worth what? $2'000.00 worth of 1TB drives? Even less? More like $1'700.00 or so -- and that's for brand new drives, faster, better, more reliable, modern technologies, SATA, etc etc etc. Power consumptions too.
By the time you're done connecting all of these, and powering them, and cooling them, and dodging the broken ones, and finding a good use for it, and controllers to run them all, I can't imagine you'll be saving many dollars for storage, if any. Not to mention your time -- although it would be fun to spend.
So in the end, you'll have all of the benefits of a massive raid solution, but it'll be expensive to build, expensive to run, and take up a rediculous amount of world space (the real storage requirement).
I don't think they can compete as functioning hard drives. I think you should use them for some other purpose -- like art, or coasters, or to hold up your table.
For example, I have about 500 issues of national geographic from the 80's. They even have those file volume collection thingies so ten get held tegother as a set. I have some rediculous number like 50 sets. These things are totally useless to me -- unlike my nintendo power issues from the '80s that my mother sold about fifteen years ago -- so I got a piece of nice glass, and now have a coffee table that sits on these magazines instead of on legs. It's a nice piece of furniture from which you can reach in a pull out a blast from the past as you sip that coffee.
http://www.youtube.com/watch?v=90jOYSfqG9I
Why not just use the largest 10 or 20, and leave the rest of 'em in the closet for now?
Either your 10-20 drive pilot project will be a raging success, and your boss will be beating down your door to get the other drives plugged in, or it'll prove to be a huge waste of time, in which case you'll be glad you didn't bother with the smaller drives.
Generally speaking, time alone will not reduce the strength of a permanent magnet. Heat, vibration, magnetic flux, and other forms of energy exposure can weaken permanent magnets. But in a box in your cabinet they are unlikely to encounter any sufficiently strong energy source to have a significant impact.
I saw this a while ago, but never got bored enough to try.
http://www.youtube.com/watch?v=fp4jQNa_9sY&feature=related
A lot of our donated computers don't come with hard drives, so we're always in need of hard drives more than just about anything else.
We wipe all drives to DoD standards before ever putting them in anything, too. (Well, anything other than the machines we use to wipe 'em.)
If you don't want to ship them all the way to Eugene, there's lots of other charities that do the same kind of thing, and probably have the same disproportionate computer to hard drive donation ratio.
If the masses can keep you down, you're not the Ubermensch.
Most SAN's aren't built with IDE/SATA drives, they are generally built with fast Fiber Channel or dual port SAS drives so that they don't have a single point of failure even in the backplane. The applications that are hosted on most SAN's care about I/O's per second as much as they do storage space. If you don't care about performance then there are cheaper solutions like a Sun Thumper or an HP DL320s that get you pretty good TB/$ while still being more reliable then most DIY whitebox storage projects. I have plenty of storage on SAN, medium speed direct attached storage and on a couple DL320s's, you use what's most appropriate for the job at hand if you're doing your job right and your management allows.
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
Sorry but if you are serious your steps should be.
1. Call a recycler and dump the drives. smaller than 200GB (keep the largest ones to give out to other employees for their home systems)
2. Buy 2 or 3 1TB HDD's
3. Install them in a box.
4. Done.
Start with the shear cost the additional equipment, then add in the cost of the electricity to run the drives and their controller. then add in the cost of HVAC to keep the room they are in cool. Will by far exceed the cost of 2 or 3 1TB drives. Not to mention the cost of your time to build, deploy and maintain.
In short. Nothing you can do with these drives will save your employer money. However proper recycling might bring in a buck or two. Not to mention the good will when you hand the largest drives to fellow employees to use at home.
I'm sorry, I'm to tired to be witty at the moment so this message will have to do.
Average HDD power consumption varies but 5W at idle and 8W under load is probably a reasonable average.
500-800W to run 100 HDDs. Some PCs use that much alone. Even these days, it's still worth using older HDDs, because the cost of replacing them with bigger and more energy efficient ones is still not low enough to cover the cost of running an older drive for a few years. Especially if your NAS supports power saving.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
One hundred drives, drawing 10W or more each (older drives were a bit more power-hungry, nowadays they're a bit under 10W) makes for 1000W. At $0.10/kWh that's $876/year. Add the power consumption of the other hardware you'll need to attach them to, and you'll surely be over $1000/year in energy costs, not to mention the purchase cost of said hardware.
:j
You said 100 drives ranging 20-300GB... that doesn't tell us much about the total capacity, but let's say it's 10TB. A terabyte disk costs less than $200 these days, a 4-port SATA PCI card can be had for $40, so with two of those and the 2 SATA ports on any cheap mobo you have a system that'll serve up your 10TB for $2000, two years of just the energy cost of your 100 disc system.
And that's not counting the headache of building your 100 disk array, the maintenance cost, and the reduced capacity due to inevitable failures with such a large number of older discs.
In short, although a cool project in theory, in practice it's not worth it today. A few years ago it would have been, but the price of storage has just dropped too steeply in the last couple of years.
I work with a group that "recycles" old machines in a developing country to provide access to young people who couldn't afford it otherwise, and even here, with free (donated) hardware it's hard to beat the falling price/performance curve of computer hardware these days. Although we could use your discs... discs (and memory) are shortest in supply. If you want to donate them to us, drop me a line.
Double that.
You need the fans, you need an extra controller card for every 4 of them, the mainboards, etc.
HI O WISE PRINCE. WHT TOOK U SO DAM LONG?
We refurbish computers and put them in the homes of low-income people, nonprofits, churches, senior centers, etc. We always need drives, and late-model computers to keep our refurbishers busy. We are a nonprofit and feel that this is an important way to bridge the digital divide.
I don't know where you're located, but we would love to have those drives, and will wipe them to Mil-spec and reuse them. that keeps them out of landfills (good for the environment) and puts good computers into the homes and tech centers of low-income communities (good for our communities and your kharma). We'll pay shipping if you would like to donate them to us.
Check us out on the web at www.ReliaTech.org. and give me a call at 510 236-7000 to discuss donating those drives and/or computers.
By the way, that donation gives you a tax deduction, too.
thanks!
Ben
You can get a bunch of firewire to ide bridge boards, and run scsi over firewire.
Keep in mind this will be noticeably slower than native ide once you get more than a certain number of drives on a single bus, but for some applications, fast disk access isn't as important.
Technically speaking, you can use USB for this too, however there are many more downsides.
Many times slower than firewire, due to the method usb uses to communicate bidirectionally.
Its not that much cheaper, and also you cant use nearly as many drives per bus.
As an example, try http://www.fwdepot.com/
Their prices are a bit high i admit, but you can build a shopping list there and look around for best price.
4 BUS firewire cards. Note that a 4 -port- card is not at all the same. That will be one bus, with a 4 port hub built in. The less drives on each bus, and the more buses you have, the more bandwidth is available to each disk, and the speed up is exponential.
One bridge board per hard drive, a few hubs and some cabling, and spread them out over your few spare pcs.
Then run something like http://evms.sf.net/ to cluster the machines together and create one giant pool of storage space out of all the drives over all the machines.
It's probably as cheap as possible for getting use out of them storage wise. Any other 'better' solution will cost a lot more too.
Of course, useful for storage and just plain useful are two different metrics.
A lot of others already mentioned donating them.
Just remember to hook 4 up at a time to a spare pc and run a good HD wipe app like http://dban.sf.net/
But there are many options to get rid of them to others with.
Charity donations for a tax write off, local community projects in need of hardware, friends, family, stocking stuffer for the staff, make a craigslist post and offer them for free (or next to), buyer comes to get it or pays shipping, do the ebay dance, etc etc
Make a bunch of chess sets out of the various parts.
Something like this.
http://www.novica.com/itemdetail/index.cfm?pid=121771
The platters of could serve as the white squares maybe?
Disk based backup solutions are worth the effort, so I can see why you're leaning this way. Unfortunately, trying to utilize ~100 PATA drives for this is going to give you nightmares for ages. Find a way to reclimate them for cash, either directly or indirectly. Hell, you can donate them to charity for a tax writeoff if you like (just make sure you DoD wipe the disks first). Take the reclimated capital and buy yourself a new data-deduplicated VTL, or a NearStor, or similar. Backup solutions need to have some level of trustworthiness to be useful, and I doubt you'll find that in a pile of aged PATA disk.
The standards for data sanitization is more stringent now. Anything that is more sensitive than Classified, and leaves the control of the organization disposing of the drives, needs to be either put through a degauser, chopped up into tiny pieces, or turned into slag. If the media is simply going to be re-used with-in the organization then wiping is okay.
SAN drives are also generally smaller in capacity, around the 250 GB range, and a 2.5" form factor. The reason for this is to maximize the spindle count. For example, having five 250GB drives with RAID 5, each running at 10,000 RPM can handle a lot more I/O (especially random seeks) than one 1TB 3.5" SATA drive. SAN drives also have a lot more cache. The reason that a most SAN companies have moved to 2.5" for their drives is because even though the platters have less capacity, the data is faster to get to on a physical basis, and case engineers have more room to engineer around the drives, especially in 1U rack enclosures than 3.5" drives.
Enterprise drives are definitely more expensive, but in this case, one gets what they pay for -- a lot more speed (especially with large, random seeks), and decent redundancy. The drives themselves are in the million to 1.4 million hour MTBF range, while consumer level drives, either don't have a rating, or the MTBF is hard to find, so the best guess is 250,000 to 500,000 hours, although some drives do have a million hour MTBF.
The key is to figure out the task at hand, and one's budget, and decide that way. Some tasks, just hooking up drives to the motherboard and using software RAID is more than workable. Other tasks are so time dependent that one has to have full hardware RAID with as many low-capacity spindles as possible to distribute the I/O far and wide. This is why Flash drives are making a good dent in the enterprise RAID market -- they are not perfect, but there is zero time wasted waiting for the head to move, and the right sector to float by.