Terabyte Storage Solutions?

DeMechman asks: "As many on Slashdot may know, storage is one thing which you can never have enough of. Given the current situation with CD/DVD rot (Personally I can attest to a 10% attrition rate) hard drives in a RAID configuration seem to be a better and more economical solution. If you own more than fifty CD/DVDs, it can be a daunting task to find a file. I am wondering if anyone has found a hardware solution that can inexpensively be set up to handle 10 or more 250GB HDDs in a RAID configuration. Primarily, has any case manufacturer tackled this niche market yet?"

In RAID, IDE has the disadvantage...

[evilviper]

It's almost funny... Once you start talking about RAID with more than 2 drives, IDE is at a disadvantage.

I'm not referring to performance, reliability, etc. (although those are serious issues), but about price.

If you have a master & a slave, then you reduce performance... That can be a very serious if you have a RAID configuration. So, if you want to put 7200RPM hard drives together, you start to need a 6 or more channel RAID card (whereas a single channel SCSI RAID card would work fine). And guess what? Decent quality 6+ channel RAID cards are very expensive, perhaps even negating the savings from using IDE drives rather than SCSI in the first place.

Remember, that's based on price-only... I haven't even begun talking about how much worse the performance would be, or reliability issues with using inexpensive IDE drives.

Warranty sucks

[Jesus IS the Devil]

Only comes with a 1 year warranty? I'm sorry, but if I'm gonna be spending over $1,000 for a storage solution, it better come with a 3 or 5 year warranty at the least. Heck all of the new Seagate drives come with 5 years warranty!

Re:Terabyte Storage

[lawpoop]

With 6 HDDs and all the other devices, what wattage power supply do you have?

Can anyone give me a rough formula of wattage/# of devices?

Re:What's "inexpensively"?

[kcarlile]

Sure, if you're just begging to lose data. Those things are 4 250 GB drives RAIDed together in a stripe. One drive goes out, all of your data is toast. Plus, there's no cooling. 4 high performance drives wedged together in one box without a fan. BAD idea. Sure, you could mirror them. Still doesn't sound very safe to me. Add in LaCie's, um, legendary customer service and reliability, and, well...

Re:What's "inexpensively"?

[leapis]

I own a large collection of DVD, and decided recently to price some large-volume storage so I could have a digital backup of my discs. What I figured out was that a 250 GB hard drive currently costs about $200 [bhphotovideo.com]. This works out to about $0.80/GB. Your average DVD contains about 7 GB of data, so you can figure a per-disc storage cost of $5.60 per disc. Based on these numbers, you can store about 35 movies per drive, so if you happen to have a couple hundred discs, you'll need at least seven drives for a RAID5 solution. Go ahead and throw in $500 for a SATA RAID [hypermicro.com] card, another $450 for a case with 7 hot swap bays [servercase.com]. And then you have to build the rest of the machine. If you spent $300 to do so, your total cost is $2650. Divide this by the total storage capacity (233 discs), and your net storage cost is $11.30 per disc. Most movies can be acquired on eBay in perfect condition for this amount or less, and you don't have the ongoing expense of also replacing drives when the die.

Obviously, these numbers are quite variable variable, and you could certainly use cheaper parts, but there is an absolute minimum cost for everything here. My conclusion was that until there is a fundamental change in the world of mass storage, in either techology or cost, this is just going to have to wait.

Re:What's "inexpensively"?

[Anonymous Coward]

"If you own more than fifty CD/DVDs, it can be a daunting task to find a file."

What's wrong with this statement that it invites derisive (if informative) giggles? I don't think the author has just 50, it's just that over 50 and finding files becomes more of a task. I'm sure some good file management databases are in order, but anyway optical disc media sure seems to fail a lot more than makes it safe for easy backup.

I have 1.5 Terabytes of personally collected data and I don't understand what's wrong with being curious about solutions over 2 TB.

Perhaps the biggest problem in true data backup is getting reliable and redundant copies off site and off the power grid. Certainly power spikes can be protected against and quality power supplies can be used, but if there's a problem with a power supply it could cause loss of data on all hard drives in one box all at once. Tape backups and optical media doesn't always work upon a restore attempt.

I can't wait for carbon rods.

Easy and cheap

[bozoman42]

I use one of the original Antec Performance cases with a door. With no external 3.5" drives (such as a floppy drive), I have 6 internal drive bays. At the moment, I'm only using one of the 4 5.25" drive bays, but you could use something such as two of PC Power and Cooling's BayCools to house another 6 3.5" drives.

Next, I'm using a 3ware 7810, which is an 8 port PATA/100 RAID controller. I'm currently using 7 ports with 120GiB Seagate drives for 3/4 a terrabyte of storage. (I want to put a hot-spare on the 8th port when I can afford it.)

Including a beefy Antec power supply, the case, the drives, and the RAID controller, it all comes in in the neighborhood of $1000. (Don't forget to add in mainboard, CPU, RAM, Network, etc. in your calculations.)

Overall, I'm extremely happy, although being as how it's PATA it's a bit cramped in their with all the cables, but I've done some work ensuring there's no excessive amounts of ribbons around the case. Likewise, I've mounted thermally controlled fans on exhaust and high-range stealth fans on intake everywhere so things keep reasonably cool. However, while the Seagates I chose for low noise and despite adding some sound dampening as well, it's still not a quiet box.

Re:It's not RAID, but ...

[Epistax]

Ah this is when their terminology really starts hurting us.

1 terabyte = 1,000,000,000,000 bytes

Try 1024^4 = 1,099,511,627,776.. wait, where'd my 100 gigs go?

Due to the exponential nature this little white lie hurts a bit more for every increment, here sacrificing just about 10% of the storage. I'm surprised they don't say 1000 gigs just to dodge the 10% mark.

For those who insist that tera means one trillon for bytes, I reference
Here [techtarget.com], here [sharpened.net], here [ic.ac.uk], here [thefreedictionary.com], here [wikipedia.org], and how about here [reference.com]. Now I'll admit the wikipedia entry has the trillion byte definition, but they basically said it is used in storage advertising.

Re:Terabyte Storage

[eo]

Anyone have any ideas on how to back up 1TB in a home environment? i.e., not $3000 tape drives & $200 tapes

I likewise have lots of data ... over 1TB in my home+office environment. Long ago, I started doing tiered backups since I couldn't afford to buy a backup system that was fast enough to maintain multiple copies of all my data. So I categorize things into three different priorities.

Priority 0 is work in progress and critical data (financial, personal, etc), which is backed up daily and rotated offsite ~weekly. I just finished transitioning from DAT to DVD-R for this.

Priority 1 is my music library and other media files I've collected over time. This is backed up incrementally daily, but doesn't change much (so there's usually not much to do in 1 day). I back this up to removable hard disks.

Priority 2 is archived work and other stuff, or files that I *could* live without (I've actually lost priority 2 data due to HD failures). I periodically go through this manually and archive things (and then take them offline). The only disadvantage here is that I have to move the archives forward to new media every so often, which usually further forces a reduction in data. ;-) It's like moving to a new apt/house -- you quickly find out what's valuable to you.

At first I thought it would be hard to choose which priority to assign to which file, but picking priorities levels carefully helped immensely. The upside is that I have backups of everything I care about, and offsite backups of really critical stuff.

Anyway, hope this helps....

Re:Easy these days.

[Anonymous Coward]

You might also want to take a look at rsnapshot [rsnapshot.org]. It implements rsync snapshot backups, and is very easy to set up and administer.

Re:What's "inexpensively"?

[BlackHawk-666]

Hard drives have approached, and may have by now passed the cost per megabyte of tape. Then again, why does the storage have to be online, why not just get a disk caddy system and start to offline that stuff. It's going to be cheaper and probably not much less usable than keeping it all online. If you want to do it on the cheap (like me) rip all your media to a couple of big drives, then store your original media very carefully away. It will last nicely when well looked after.

You may have to repeat this process every few years, like I am doing now since I bought an iPod, and found all those neat OGG files to be less than playable on it. This seems like only a moderate hassle for the gains of storing the media compressed on a server, thus having it constantly available, and yet still small enough not to require terrabytes of storage.

Our Solution

[Pathway]

My school asked for me to create just a solution. Here's what I ended up with:

* 4U Enlight Case.
* 2 5-drive SuperMicro Hot Swap removable Drive cages. (SATA)
* 3Ware 12 port 8506.
* Supermicro Dual Xeon Motherboard.
* 2 Xeon 2.4Gig Hyperthreading Proccessors.
* 10 250 Gig SATA Harddrives.

The 3Ware card has a limitation of 2 Terabytes for a single volume, so we used a 9-Drive Raid 5 with 1 hot spare to make our large drive.

We used Debian Sarge, and BackupPC to backup our school's servers. We can backup EVERYTHIHNG now.

Oh, and this solution cost us less than half of the pre-designed solutions. The school has been very happy with it indeed.

Re:Many have

[jonbrewer]

I think the Apple is the cheapest any sane person or organization would go. They have engineered a solution, as opposed to assembling a solution as you propose. All sorts of things can go wrong with such an assembled solution, such as heat, vibrations, power fluctuations, drive failure, data corruption, etc. The Apple solution takes all these in to account, and does so at a reasonable price. Better yet it can be replaced or repaired at any time, as opposed to an assembled solution, which needs the original assembler to be present (not fired or on vacation) if it breaks.

Re:What's "inexpensively"?

[gujo-odori]

UPSes and redundant power supplies are great, but as the grandparent metioned, a bad power supply can corrupt your data. That's true even in a redundant-power machine.

A friend of mine once lost all the data on two drives (RAID 1) in a country with extremely reliable power (Japan; even during typhoons I never once had a power outage in 8 years) when the UPS suddenly died one day and dumped the whole battery load into the computer. The white smoke escaped from everything.

If your data is really valuable, offline storage is not a luxury, it's a necessity. Get a DLT drive (or a changer, if you can afford it). Offsite is easy. Keep at least one backup set at your office. If your house burns down, you're covered. If something so bad happens that it destroys both your house and your office, you have bigger problems than the lost data :-)

Re:It's not RAID, but ...

[threephaseboy]

I don't think the fibre channel controller is that much of the cost. Pricewatch lists FC PCI cards for ~$105. Probably the same cost as FW800,ethernet, and the related embedded controller those would require. The entry level xserve raid is $5k for 1TB (4*250G), I don't think that price would be cut too much by merely taking out half the drive bays. Possibly some would be cut by taking out the redundant PSU and raid controller, but probably not more than a couple hundred. PSUs are cheap, and full raid5 SATA 4-channel PCI controllers retail for a couple hundred, certainly less in OEM quantities for just the chipset.
If all you really want cheap bulk storage, what you probably want is the Lacie Bigger Disk [lacie.com], 1.6T for $2.2k, FW800. Add on a cheap linux box with gigE and FW800 for $300 or so, and you have your NAS with more storage than the xserve raid, more connectivity for half the price. But I wouldn't put it against the xserve raid for reliability or performance any day.