Turnkey Linux RAID Solutions?

Total-Gig-Age asks: "I want to buy an expandable RAID system for home storage of large media files (music, film, and photo). I'm absolutely unwilling to rely on optical discs (bit rot, not always online) and un-RAID-ed hard drives (unsafe: if it fails, you're screwed). The thing is, I don't have time to shop for and configure a RAID system myself, and I want a turnkey solution that will just work out of the box. I'm aware of Apple's XServe, but $6000 for 1 TB is just too expensive. What are my best options if I want to buy an open source system that I can maintain and upgrade if need be? Any recommendations on a full set of components, so that I don't have to spend a week shopping? Trustworthy online companies? Can I trust a local store to do it for me? Is it better to keep the server as a separate machine? Finally, how much should I expect to spend if I want something that doesn't suck (for 1TB say)? I can find plenty of info on how to set up RAID on the Internet, but I just want to be told what to buy so I can get on with other things, even though I could probably handle setting the whole thing up myself if I had to."

Comments

You know why it costs more, right?

[beegle]

You want good quality, but you don't want to pay for it. Ummm, right.

Quality comes at a price. Everybody learns this eventually. With a DIY solution, the price is your time. You can make something really great if you're willing to burn a weekend or two on it.

If you're buying something, you can have "moderately expensive, stable, and really limited", "really cheap, but likely to fall apart or catch fire", or "really expensive and really flexible".

The other thing that you run into with a sommercial system is the difference between home and business requirements. For a business with a machine room, dust, humidity, and temperature are easy to control. A noisy unit is fine. Under your desk, temperature and dust build-up will be a problem, and the thing'll sound like a jet engine.

Promise UltraTrak SX4000

[Yonder Way]

I got a couple of Promise UltraTrak SX4000 4 Channel External RAID Enclosures at $WORK to evaluate and while they aren't fast enough for me to move a production Oracle server to, they are plenty fast for a home media server and are quite affordable.

Each enclosure presents itself as a SCSI drive. You can chain enclosures together and use LVM to make them available to one (or more) filesystem(s). Hardware RAID takes place within each enclosure, but not across them.

Mine were set up with four 300GB drives which, after RAID5 overhead, gave me 900GB of usable space per enclosure (I had two enclosures). You could easily use 400GB drives and have 1.2TB per enclosure if you wanted. I would think that total cost for the enclosure with 400GB disks would be around $2500 US.

Re:Promise UltraTrak SX4000

[aminorex]

Take a generic low-end PC (say, sempron 2200, with a pc-chips or ecs motherboard, 128MB DDR -- it's all overkill at this point) in a tall, well-ventilated case with at least 300W power, for about $150, and add 4x 250GB IDE drives, for about $450. Run as raid 5 on a debian sarge install. That gives you 750GB of online storage suitable for streaming media for about $600. When 400GB drives fall to the same $0.40/MB price point, make a new box and sell the old one on ebay, to pay about half the cost of the upgrade.

Re:Promise UltraTrak SX4000

[tweedlebait]

I've had a horrible experience with 2 ultratracks-

The controller decided to flag 1 drive in a raid5 as out of sync, (the drive was perfect) then in a few mins. bring it back online and decide that all other drives were out of sync and needed to be rebuilt based on the 1st drive it just de-synched. All the while happily reading the 1st drive's data incorrectly.

On both accounts (after much pressing) promise engineers said the firmware immolated itself. This was all during a few months of uptime on the first occasion and 1 week on a different (test) system and different model of card.

I moved to 3ware since then (about 2 yrs ago) without any real problems. Losing a TB in a redundant system is hard to explain to anyone.

Re:Promise UltraTrak SX4000

[aminorex]

I agree with everything you said, but would note
that by adding PCI RAID IDE controllers, it scales
to the available case cooling and mount points.
I'd probably add some mounting brackets from
erector set pieces and a few fans (and maybe some
acoustic insulation to my closet!!) if I wanted to
scale up to, say 12x400. But that really is the
wall for my personal taste, 3.6TB of RAID 5.
If I ran out of erector set pieces, I'd just pick
up an old gutted tower server case for the job
from ebay or axe-man.

I'd peg the 3.6TB solution at 550 for the
platform, including 3 promise tx2000 cards, and
just under $4k for the drives. That's $1.25
per GB, compared to $0.80 per GB for the 4x250
model. Personally, I'd rather keep multiple
servers, say one for anime and music, one for
data files, one for movies... With network
cross-mounting, it all looks like one big
directory tree anyhow. Of course the middle
road, more drives at the lower $/GB 250GB, $113
sweet spot, still lets you scale to 2.25GB, and
keep it close to $0.80/GB. And it is easier to
upgrade incrementally, by moving one RAID set at
a time to large physical drives, since shuffling
data between controllers is a lot faster than
pumping it over a LAN connection.

I find it easier, faster, and vastly cheaper, to
stick lots of drives and controllers and fans in a
big tower than to buy boutique hardware. I also
find it more reliable, since my personal budget
allows for plenty of spare hardware at the lower
pricepoint. If you really want to run multiple
RAID controllers in a pc case, though, remember
to pick a mondo power supply. you'll probably
need a bunch of power Y connectors too, unless
you're using all SATA drives.

And when the house burns down???

[linuxwrangler]

Sure, RAID helps - go grab a 3Ware card for your machine and mirror your data. But things like RAID and dual power-supplies are really in their element when system-availability is important. Ths system keeps running and you can hot-swap the drive or schedule off-hours down-time.

For keeping your data safe, however, RAID is mostly useless - something you will come to realize when the house containing your RAID burns down or when the RAID is stolen by burglars or a human/software glitch manages to "dd if=/dev/zero of=/dev/yourdisk" your RAID. Power-surge, tornado, flood...the list of things against which RAID fails to protect is long.

If you really care about your data you _must_ perform regular backups and take them off site. I rsync my photos to my work machine and use a VXA tape drive for regular backups.

And I don't bother with RAID except on servers at work.

Re:And when the house burns down???

[walt-sjc]

I'll second 3ware. I've had problems with Promise and their "fake" raid, changing chipsets ever 5 minutes and breaking compatability, etc.

3ware isn't cheap but you get what you pay for. High-quality, great performance, great support, and it "just works" with Linux.

I've had issues with several brands / chipsets of onboard SATA controllers, which is why I went with 3ware. Software raid is all fine and good if you can get reliable drivers! If not, then avoid it.

With 1T of storage, you are talking quite a few drives. Make sure you have good cooling, big power supply, etc.

Re:And when the house burns down???

[Daedala]

Perhaps the cheaper RAID would enable him to afford a good tape backup system and offsite storage.

$6,000

[vasqzr]

The thing is, I don't have time to shop for and configure a RAID system myself, and I want a turnkey solution that will just work out of the box


You're paying $6,000 because Apple does the work for you.

I suppose you could get a tower PC, and fill it with hard drives and setup RAID. Cheaper? Yea. Reliable? Yea. But there's more to it than that.

I googled for 'building 1TB server'

http://www.martinandalex.com/blog/archives/2005/01 /building_1tb_ra.html

Home 1TB RAID Server
CPU Athlon 3200+ $199 Frys 11x multiplier, should over clock to 2.6GHz easily
Memory 1GB Corsair 4400C25 $275 Very fast at DDR466
Motherboard ASUS K8N-E Deluxe $149.99 Frys, 6 SATA RAID chips on Motherboard, 3GB memory
Case SUPERMICRO Beige 4U Rackmount Chassis, Model "SC742T-550 Beige" $307.50 New Egg. Has 7 SATA backplane built in
CD drive NEC 3500A $67 newegg or zipzoomfly
system drive WD740GD $185 10000 rpm system drive
Data drive Maxtor DiamondMax 10 250GB $149.99 ($.59/gig vs $.68/gig for 300GB) 5 of these bad boyz

total: $1933.48 or less than $2 per gig for RAID. Half the cost of white boxes and 1/3 the cost of anything from the channel.

Here's another article, more information

http://www.ethics-gradient.net/myth/storage.html

Re:$6k/1TB too much?

[DShard]

Considering that a sata card with eight channels is roughly $500 and 6 $200 gig drives can be had for less than a grand I would say that for a home/small office $6k is to expensive for what you get. considering file/print services can be handled very well by the cheapest processor on the market, a total system can be had easily for half that with full support. 4TB is a different situation since you need a lot of io channels and I would think that, realistically, $6k is too cheap.

OT: definition of sommercial

[Short Circuit]

sommercial - adj - The business of psychoactive drugs.

sommercial - n - An advertisement for soma.

Source: A Brave New World, by Aldous Huxley

This is not a stupid question...

[egarland]

Whenever a post about inexpensive Linux based RAID storage comes up on slasodot there is a flurry of:

• "You know why it costs more, right?"
  
• "if you want easy, you pay for it, if you want cheap, you work for it"
  
• "if it's important, then $$ shouldn't matter"
  

and other nonsense excuses for not answering the question.

I'm guessing people have spent a LOT of money on reliable storage solutions and tend to be irrationally dismissive of the possibility of inexpensive redundant storage.

The fact is, if you know Linux well, maintaining a Linux based RAID array for home use is perfectly reasonable and generally quite painless. I build an inexpensive 4 drive 480GB RAID array a few years ago that I've been delighted with since. I have survived a disk failure with minimal downtime and no data loss.

"And when the house burns down?"

I'm so tired of this stupid argument. Data loss due to fire will happen with or without RAID. The fact is, losing a disk is much more likely than having your house burn down by a very large margin (I'd take a rough guess that disk loss in a 8 disk system is about 10,000 times more likely than disk loss from fire). But even if they happened with the same frequency you'd still be reducing your exposure by 50% by eliminating data loss from disk failure with RAID.

I have yet to find an online company selling properly configured systems for a reasonable price.

I thought about building a standalone storage server recently and saved my design in a newegg wishlist

For rack mount RAID systems I like the design cases they have at www.rackmountpro.com but I've never dealt with them personally so I can't say how well they work.

LaCie F800

[kinema]

LaCie recently announced their F800 desktop RAID solution. 1.0, 1.6 and 2.0 terabyte models are available. The array connects to a host computer via IEEE1394 (Firewire) or USB 2.0. The array can be configured as RAID levels 0, 0+1, 5, and RAID 5+hot spare. PriceGrabber lists the 1.0TB version for just under US$1500 and the 1.6TB for US$2700.

mmmm

[Fr05t]

mmmmm turkey *drool*.. oh you didn't say turkey? :(

SAN?

[Dr.Dubious DDQ]

I'm looking for a roughly similar "solution" - adequate performance with redundancy and lots of storage space. A "SAN" ("Storage Area Network" - one of the current buzzwords going around these days) might be useful.

Either iSCSI (if you want economical and standards-based) or Fiber-Channel (if you're wealthy and the speed of writes to the hard drive array is critical) based boxes of hard drives seems to be an option, and from the point of view of the server (or whatever computer is using them) they are just another hard drive. Or so the materials I've read say. (Think of them as an external RAID box...)

iSCSI seems to be limited to 1Gb speeds (unless you can get your hands on 10Gb ethernet cards and switch, which I gather are now available), which to me seems perfectly adequate for most file-server type uses. It looks to my still-new-to-the-area eye that you can also do a lot of potentially useful tricks because of the standard IP-based nature of the data transfer (such as being able to mount a "hard drive" directly over the internet or a LAN, if you have some reason to need to do so). Fiber-channel is faster (2Gb seems to be typical, 4Gb is apparently getting fairly established, and 8Gb is available if you're made of money) but requires specialized and fairly expensive hardware ($500+ for each fiber-channel interface card at the LOW end, as I recall, plus several thousand for the fiber-channel equivalent of a "switch".)

On the subject of iSCSI, there seem to be active projects with both "target" (iSCSI device server) drivers and "client" (iSCSI device mounting) drivers for Linux on Sourceforge...

Corrections welcome, of course...

Incidentally, that's not to be confused with "NAS" ("Network Attached Storage") which as far as I can tell is a buzzword used by people due to the fact that "file server" doesn't sound "cool" any more...your "NAS" might be using a "SAN" to store the drives that it is serving...

In any case, this may be me trying to "hijack" this Ask Slashdot, but what do people here think of the "SAN" concept and its implementations?

Learn from others who've done this already

[cowbutt]

a good recipe for a 1.2TB Linux-based RAID for under US$1600.