• We should now be fully online following an overnight outage. Apologies for any inconvenience, we do not expect there to be any further issues.

Anyone here ever built their own RAID from scratch?

Arkitech

Diamond Member
Apr 13, 2000
8,356
4
76
I would like to build my own RAID for as cheap as humanly possible. I don't know much about RAID's but with hard drive prices getting cheaper I think it's time to look into investing in a nice backup solution. It would also be nice to have all of my data in one place instead of on multiple PCs and various usb drives.

It would be really cool to have the ability to include mobile racks like these (http://www.monoprice.com/produ...d=3857&seq=1&format=2) as well.

So is it possible to build out a cheap RAID for under 100 bucks?
 

MalVeauX

Senior member
Dec 19, 2008
653
176
116
Heya,

For starters, it's time to read up on RAID and HDD's in general. RAID is NOT a backup solution. Just get that idea out of your head and save yourself the time, money and data loss that you may encounter. RAID is not for backing up, it's for increasing capacity (back when drives were truly small), increasing performance (this we still can get a lot out of) and mainly for Uptime (uptime is the ability to have the machine on, serving data, all the time, or at least as often as possible even if a hardware failure occurs).

Based on what you want, it sounds to me like what you need is a NAS (network attached storage). And I recommend against RAID in your NAS machine unless you're willing to build a monstrously huge multiple-terabyte-RAID5 array. My reasoning is, you will outgrow the size of your RAID array over time. And when you do, you can't just add a drive to it and increase capacity again. You have to rebuild, which means you have to move data. You will be in a big load of trouble trying to move 1+ terabytes of data off an old array, onto something else, while you rebuild the array--meaning, whole new drives, or other media. This is really, really hard, without having basically a whole new RAID to accept the data from. This is why I suggest you avoid RAID in a NAS. Instead, just buy the largest capacity drives you can that are reliable (the 1 tb drives are very good right now and cheap) and have each one independent from the other in a NAS. Very simple. Add drives as you need them. Access all these drives & data sets from ANY machine on your network.

Again, RAID is not a good backup solution. It's also not a good NAS solution (unless you need uptime reliability and then RAID5 and RAID6 are ok, but this is major stuff, not something you ever need at home).

And as for building a NAS, yes, you can do it for $100 (minus the cost of drives of course). Though I would try to put more into it to have a good reliable power source and motherboard to work from.

Very best,
 

RebateMonger

Elite Member
Dec 24, 2005
11,586
0
0
Consider scrounging up an older PC (1 GHz, 500 MB RAM) and installing Windows Home Server ($100) on it. That will let you toss in a variety of hard drives and see them as a single drive. As long as you physically have a way to add a drive to the PC, you can make it part of your drive pool and store data to it.

You can also enable disk redundancy, so that the failure of a single hard drive won't cause loss of data. But if your data is really important, it's safer to make back ups data to a separate drive(s). Don't trust important data to any single hard drive or single RAID array.
 

shempf

Member
Dec 7, 2008
74
0
0
WHS or unRaid would be my software choices.
Sounds like you might just want a file server to backup (via imaging imo)
 

Madwand1

Diamond Member
Jan 23, 2006
3,309
0
76
It's easily doable, but not for just $100, and as RAID alone is not a backup, if you're looking to accumulate tons of personal data on such a machine, you take the risk of losing everything at some time, making the entire effort worse than bad. To mitigate that, you need to plan at the onset for a real backup, which doesn't quite, but almost doubles your cost off the bat.
 

BonzaiDuck

Lifer
Jun 30, 2004
16,636
2,029
126
It depends on what you want, and what you want to pay for, as well as your tolerance for a slightly higher power bill over the long run unless you can run the system with sleep-mode enabled such that recovery from sleep doesn't mess with your clock settings -- even the memory timings.

Just by its very nature, large volume disk storage is a bottleneck in the pyramid of system architecture. At the top, CPU registers are extremely fast, and store minute volumes of information. At the disk end, the devices are relatively slow, and store large quantities of information at a low price ratio.

It makes sense to widen the bottleneck, if bus speeds are ample. With RAID0, you get performance increase to the limit of double throughput, but the chances of hard-disk failure are proportional to the number of disks in the array, while performance increases with the number of disks, as does power consumption. If one disk fails, you lose all your data -- effectively, as stored on the entire array.

With RAID1, you get data reliability with some small increase in performance improvement. You consume more power in measure to the number of disks, and data storage capacity is confined to single-drive size.

There are combinations of RAID0 and RAID1, but they effectively multiply your power requirement because they use more disks, and the capacity of the array for storing data is limited to the RAID0 component.

With RAID5, you need at least 3 disks. Your storage capacity at that threshold is 2/3 of the total for all disks, for the increase in parity data about the data that is stored on any remaining two drives. Performance compared to RAID0 is somewhat degraded because of need to process the parity data. If you increase the size of a RAID5, you get proportional increases in both speed and capacity, but you increase the chance of two or more drives failing at once, and RAID5 is only tolerant of a single drive failure.

RAID6 -- check if I'm wrong -- is tolerant of a two-drive failure, and I think it needs at least four disks.

On the power-consumption end, some company -- can't remember but thought it was "ACORN" -- produced a NAS that employed various RAID configurations for 5,400 rpm notebook drives. These drives consume less power, and are smaller, so the proper configuration would allow you to do a round-robin on spares that can be stored in a fireproof safe as clones or bootable backup partitions of the array. And Western Digital told me that SATA 2.5" notebook drives are desktop-system-compatible, so you can also go that way . . . .

But ultimately, if you want an equally significant performance boost, you would want to spend $300 or more on a hardware controller with its own onboard processor/XOR-engine and at least 128MB of cache or buffer RAM.

I don't just assume my RAID5 has assured reliability; I back it up to USB drives or a networked-file-server RAID5 array. And I have hot-swap drive-caddies for backup of the server's RAID5.
 

yh125d

Diamond Member
Dec 23, 2006
6,886
0
76
Raid not a solution for backup? You know the R is for redundant right?
 

RebateMonger

Elite Member
Dec 24, 2005
11,586
0
0
Originally posted by: yh125d
Raid not a solution for backup? You know the R is for redundant right?
There's lots of ways to lose data on a redundant RAID array. I've seen it happen more than once. It's MUCH harder lose significant data with a separate backup of your data.

The theoretical reliability of redundant RAID arrays is somewhat compromised by the added complexity and increased points of failure. Additionally, a RAID array does nothing to stop data loss caused by human error, theft, fire, flood, etc. The primary purpose of redundant RAID is uptime, not backups.
 

MalVeauX

Senior member
Dec 19, 2008
653
176
116
Originally posted by: yh125d
Raid not a solution for backup? You know the R is for redundant right?

Redundancy and backup are two different things. Redundancy is to protect the array against failure if a single drive fails. That's why RAID5, 6, 1 exist. But if more than one fails on RAID5, you lose the data. Period. If you lose more than 3 in RAID6, you lose it. In RAID1, if one drive fails, the other is an exact copy. But if the data is important, if something physically happens, you lose the data, ie, if that machine gets knocked off and all the heads crash for example. The other thing is that while RAID has redundancy in some levels of RAID (not all!), rebuilding is not always perfect and not always successful. Look around the forums and other forums. Even when someone's drive failed in RAID1 and RAID5, there are people who were unable to get the other drive to work, and/or rebuild the data from parity. Why? It's imperfect. This is why RAID is not a backup solution. Backup is all about safety. And it's best when it's physically not in the same place as the original and best when it's not on a device that has moving parts that can fail (HDD's). If you do backup on HDD's, it's best to have multiple copies of the same data on several drives. Sounds expensive, and it is, but that's the only way to really make HDD's a relatively safe option for backing up data that you truly can't afford to lose.

Just because RAID can at some levels keep it's data from being lost when a physical drive failure occurs, it does not protect against data corruption and/or physical environmental issues (fire, water damage, theft, etc), doesn't make it a back up solution. And again, these levels of RAID do not protect very much. A single drive failure is the only `redundancy' you get from RAID5. Well, what if you have 5 drives? One fails, well, two can fail. Think about it. This is why it's not a back up solution, but rather, and up time solution.

A successful RAID array system for real backing up of data requires several copies of the data sets on several separate RAID arrays. It's pricey. It's exceedingly more than any home consumer should ever need. It's what businesses and massive servers do, and they do it in multiple locations. This is where the `redundancy' of RAID can actually be seen, at the really macro level, where cost is of less issue compared to actually preserving data.

Very best,
 

mxnerd

Diamond Member
Jul 6, 2007
6,799
1,103
126
Originally posted by: RebateMonger
The primary purpose of redundant RAID is uptime, not backups.

This sums it up, and MalVeauX told the details.

RAID IS FOR UPTIME. FOR NOT BACKUPS.

Sorry, I have to shout. :)
 

BonzaiDuck

Lifer
Jun 30, 2004
16,636
2,029
126
Originally posted by: mxnerd
Originally posted by: RebateMonger
The primary purpose of redundant RAID is uptime, not backups.

This sums it up, and MalVeauX told the details.

RAID IS FOR UPTIME. FOR NOT BACKUPS.

Sorry, I have to shout. :)

Ditto -- that. Redundancy -- with some RAID configurations -- means an increase in reliability against drive failure. It is not "backup." And this "reliability" goes hand in hand with "uptime." Only with RAID1 could you argue that the configuration creates a mirrored "backup," allowing a periodic round-robin of disk-swaps so that you can keep one in secure storage.

Uptime may also mean that you don't have to use a backup partition and copy it onto the boot drive or repaired array -- you may skip a step or two since with (say, RAID5) the data is still "there." This does not mean that backup isn't something separate, or that it isn't a wise practice. You could argue that increased reliability of certain configurations means that the frequency of backup is a matter of lesser urgency, but periodic backup is still necessary and wise.
 

Arkitech

Diamond Member
Apr 13, 2000
8,356
4
76
Originally posted by: MalVeauX
Heya,

For starters, it's time to read up on RAID and HDD's in general. RAID is NOT a backup solution. Just get that idea out of your head and save yourself the time, money and data loss that you may encounter. RAID is not for backing up, it's for increasing capacity (back when drives were truly small), increasing performance (this we still can get a lot out of) and mainly for Uptime (uptime is the ability to have the machine on, serving data, all the time, or at least as often as possible even if a hardware failure occurs).

Based on what you want, it sounds to me like what you need is a NAS (network attached storage). And I recommend against RAID in your NAS machine unless you're willing to build a monstrously huge multiple-terabyte-RAID5 array. My reasoning is, you will outgrow the size of your RAID array over time. And when you do, you can't just add a drive to it and increase capacity again. You have to rebuild, which means you have to move data. You will be in a big load of trouble trying to move 1+ terabytes of data off an old array, onto something else, while you rebuild the array--meaning, whole new drives, or other media. This is really, really hard, without having basically a whole new RAID to accept the data from. This is why I suggest you avoid RAID in a NAS. Instead, just buy the largest capacity drives you can that are reliable (the 1 tb drives are very good right now and cheap) and have each one independent from the other in a NAS. Very simple. Add drives as you need them. Access all these drives & data sets from ANY machine on your network.

Again, RAID is not a good backup solution. It's also not a good NAS solution (unless you need uptime reliability and then RAID5 and RAID6 are ok, but this is major stuff, not something you ever need at home).

And as for building a NAS, yes, you can do it for $100 (minus the cost of drives of course). Though I would try to put more into it to have a good reliable power source and motherboard to work from.

Very best,

thanks for the input, you probably saved me a ton of money and headaches. I recently had a hard drive crash and stupid me did'nt have anything backed up, I figure now it's time to do the right thing. Guess I'll have to do some research into a NAS solution.
 

Mycophiles

Member
Jul 19, 2005
36
0
0
Great thread. I've always thought Raid was for backup. It makes perfect sense in theory but and upon a quick explanation it is no wonder so many are fooled.

So: for those of use that are looking into RAID as 'backup' what is the answer? (when 1tb drives are so cheap) I'm in the process of building a server right now and am looking at Windows Home Server. It looks to be extremely nice. Actually I've been looking around to see if you can configure a RAID array within WHS to go along with the 'pool'.

Considering this new evidence WHS looks exactly like what you would want to backup files. EXACTLY What seems to be the case within WHS is that a backup is done of every file across partitions in the 'pool'. HD's can be varying sizes so no worries about having the exact same drive. This option is starting to look very appealing except that it's a 1to1 backup and there is no space saved such as in a RAID 5 array.

So, now that I've bought a Perc 5/i card. What do I do with it? Ohh and what is NAS? I gotta do some more research to figure out what exactly I need for this server. Just so everyone is clear. I'm building a Home Server for file/movie/music management. Just putting everything in one place and trying to secure it as best as I can.
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
raid is not a backup solution FOR THE DRIVES WITHIN THE ARRAY, but it is a dataloss prevention mechanism, And it CAN be a backup solution if you are backing a drive on computer A to a raid array on computer B.
RAID1 or RAID5 on a single computer might even be more effective in protecting your data than a single drive with some sort of backup.
but NEITHER provides full coverage. I am loathe to say that for full protection you should use "raid and backups" because the terms are too fuzzy and it can be implemented wrong. What you should do is form a data loss prevention strategy, Make a LIST of all possible data loss methods. And then make a list of possible protections against each one. And then make a combination of such methods as you are comfortable with, can afford, and feel is necessary for the type of data in question. (the type of data matters a lot!)

Because a raid1 array is so EFFECTIVE for protecting data against many different types of catastrophic loss, many people might say "for my backups I use raid" when they mean "for the safety of my data I use raid". If having to choose only one between an external drive backup, raid1 array as primary storage, and DVD backups. I would choose raid1 array... for my data.

Anyways. building raid is easy. The question is does the op mean for the 100$ to include the price of the drives. And how much space does he need.
 

BonzaiDuck

Lifer
Jun 30, 2004
16,636
2,029
126
Originally posted by: taltamir
raid is not a backup solution FOR THE DRIVES WITHIN THE ARRAY, but it is a dataloss prevention mechanism, And it CAN be a backup solution if you are backing a drive on computer A to a raid array on computer B.
RAID1 or RAID5 on a single computer might even be more effective in protecting your data than a single drive with some sort of backup.
but NEITHER provides full coverage. I am loathe to say that for full protection you should use "raid and backups" because the terms are too fuzzy and it can be implemented wrong. What you should do is form a data loss prevention strategy, Make a LIST of all possible data loss methods. And then make a list of possible protections against each one. And then make a combination of such methods as you are comfortable with, can afford, and feel is necessary for the type of data in question. (the type of data matters a lot!)

Because a raid1 array is so EFFECTIVE for protecting data against many different types of catastrophic loss, many people might say "for my backups I use raid" when they mean "for the safety of my data I use raid". If having to choose only one between an external drive backup, raid1 array as primary storage, and DVD backups. I would choose raid1 array... for my data.

Anyways. building raid is easy. The question is does the op mean for the 100$ to include the price of the drives. And how much space does he need.

Here's where I'm going to trash my . . . . "credibility" -- by telling this story. If you kept up with the news, there has been a frenzy over Rush Limbaugh. I fancy I'm a snickering sort of humorist, but I'd had an experience back around 2001 that would leave people calling me nuts. I had read through all the material Charles Manson wrote in prison after a trip to Goler Wash (Death Valley) and Barker Ranch with the SUV.

Several months later, I logged onto a forum that contained a thread entitled "Mark of the Beast," and I discovered this 16-year-old kid having a long dialog with someone who called himself "Will." I recognized the writing style, the vocabulary, syntax, usage, use of rhetorical questions in every paragraph, the mentally-loose quality of hare-brained rants -- even a paragraph in the Manson posts that used the word "will" in noun and verb form about 20 times in a single paragraph.

Will as much as admitted or hinted at his identity, and then left the forum like a vampire glimpsing daylight -- after another rant ending with some dissertation about a court-proceeding and "no smoking gun." The sixteen-year-old kid had said:

"My dad tells me not to listen to anything but Rush Limbaugh. . . . . "

I forgot to back up the forum pages and downloads of the comparison passages, so I can't even offer anything as "proof" of the "possibilities."

So now -- even if you all think I've lost my grip entirely -- REMEMBER -- ALWAYS BACK UP YOUR DATA. [More later about who killed JFK . . . ]

 

NXIL

Senior member
Apr 14, 2005
774
0
0

Why Mirroring Is Not a Backup Solution

Posted on Friday January 02, @09:25AM

from the pointed-lesson dept.
Data Storage IT

Craig writes "Journalspace.com has fallen and can't get up. The post on their site describes how their entire database was overwritten through either some inconceivable OS or application bug, or more likely a malicious act. Regardless of how the data was lost, their undoing appears to have been that they treated drive mirroring as a backup and have now paid the ultimate price for not having point-in-time backups of the data that was their business."


http://hardware.slashdot.org/a...l?sid=09/01/02/1546214

Ouch.

RAID or not, if all your data is in one location, it can easily get stolen, burned, zapped by power grid problems, flooded by a broken pipe, tornado, hurricane, earthquake, smashed by angry significant other, or destroyed bit by bit by malware/worms/viruses. Not good.

Backups should include off site backups....as noted in posts above, RAID can be helpful, but, it can be a disaster to rely on it alone.

NXIL
 

skyking

Lifer
Nov 21, 2001
22,777
5,939
146
Originally posted by: taltamir
raid is not a backup solution FOR THE DRIVES WITHIN THE ARRAY, but it is a dataloss prevention mechanism, And it CAN be a backup solution if you are backing a drive on computer A to a raid array on computer B.
RAID1 or RAID5 on a single computer might even be more effective in protecting your data than a single drive with some sort of backup.
but NEITHER provides full coverage. I am loathe to say that for full protection you should use "raid and backups" because the terms are too fuzzy and it can be implemented wrong. What you should do is form a data loss prevention strategy, Make a LIST of all possible data loss methods. And then make a list of possible protections against each one. And then make a combination of such methods as you are comfortable with, can afford, and feel is necessary for the type of data in question. (the type of data matters a lot!)

Because a raid1 array is so EFFECTIVE for protecting data against many different types of catastrophic loss, many people might say "for my backups I use raid" when they mean "for the safety of my data I use raid". If having to choose only one between an external drive backup, raid1 array as primary storage, and DVD backups. I would choose raid1 array... for my data.

Anyways. building raid is easy. The question is does the op mean for the 100$ to include the price of the drives. And how much space does he need.

I perform most of my backups to other drives myself. It is still volatile storage but on two servers in two locations, on RAID1. I will burn an occaisional stack of DVD's for stuff that is old and fairly static in nature.
I use Dirvish to perform the backups. It is easy to configure and the backups are easy to get a file out of. You can choose the retention for as long as you like.