Home Network Data Storage Device

It happened again- a machine on my home network died. Taking with it tons of data. It's mostly backed up. No huge loss. But I finally think it's time to get some sort of network raid disk. A unified place to safely store data accessible to the numerous machines on my home lan. So now I pose to Slashdot readers- what are your recommendations? I'm looking for something with RAID and SMB sharing. At least a quarter TB, probably a half, but with some room to grow. What have you used? What works? What fails?

Re:Simple answer.

[CerebusUS]

I believe Buy.com has the 1TB version (~700GB usuable in a RAID 5 config) for about $700 as well.

I just convinced my boss to buy 2 of these for backups

Re:Wow! Research!

[SuiteSisterMary]

What's wrong with somebody looking for user testimonials and advice about actual installation and use from a tech site?

Xbox does the trick

[Anonymous Coward]

Recipe for xbox raid network storage:

Add 1 Xbox ($130)
Add 1 mod chip ($30)
Add 2 250 GB HDDs ($250) -- you can either disconnect the CD-ROM or follow instructions on adding a second hard drive, but disconnecting allows everything to be internal
Add 1 Linux for Xbox ($0)
Stir in Raid 5 set-up and samba

And you've got yourself a headless quarter terabyte raid 5 network server for a low-budget of $410.

RAID != backup

[undeadly]

It happened again- a machine on my home network died. Taking with it tons of data. It's mostly backed up. No huge loss. But I finally think it's time to get some sort of network raid disk. A unified place to safely store data accessible to the numerous machines on my home lan.

RAID could help with downtime, but is not a substitute for backup, really. Tape backup is still very expensive (high inital cost), and DVD's are limited in both quality and storage capacity. Well, I use both, but then my storage needs are slight since I burn my most important data to a DVD-RAM disc every night.

What OpenBSD thinks about RAID: [openbsd.org]

RAID (Redundant Array of Inexpensive Disks) gives an opportunity to use multiple drives to give better performance, capacity and/or redundancy than one can get out of a single drive alone. While a full discussion of the benefits and risks of RAID are outside the scope of this article, there are a couple points that are important to make here:

* RAID has nothing to do with backup.
* By itself, RAID will not eliminate down-time.

If this is new information to you, this is not a good starting point for your exploration of RAID.

Re:The Poor Man's RAID Array

[timeOday]

Why go with RAID at all? Hear me out.

Whatever you do, you MUST be protected from accidental deletion and corruption. That means you need a backup, which RAID is not. Now assuming you maintain a separate backup, why waste disk space on a separate "hot" backup, which RAID (not 0) provides? If this is home use, you don't care about the downtime required to restore from background in event of a disk failure.

If you're like me, you don't want to buy a bunch of identical disks at once for home use. Instead, you have a range of larger newer disks, and smaller older disks. . This means the disks you want to use are NOT all the same size, as required by RAID (AFAIK). Instead, you can use LVM [tldp.org] with linear mapping to combine smaller drives into one larger one, even if the physical drives are mismatched sizes. Create one logical volume for live, and one for backup, and do nightly updates of the backup. You probably don't want/need to compress the backup if the bulk of your files are already compressed media files.

Re:It's Time my Son

[j-pimp]

Faster? Yes. Better? Yes. Only for Mac/Linux? NO!!

I'll skip the whole you forgot about the inventors of nfs [sun.com] whining and just point out that better is highly subjective. First of all you can do password authentication with samba. With NFS its by uid only. While thats convient if your exporting home directories where all the machines can trust each other and are running UNIX, if you want users to be able to mount or browse shares themselves then samba is the way to go. Also samba allows you to share printers and do domain based authentication. Perhaps NFSv4 can do that and you can prove me wrong, but you should have specified that as I'm sure the average NFS user is as unaware of the features of NFS4 as I am.

Re:Simple answer.

[MikeFM]

Or run it behind a Linux box and use a caching filesystem in front of it. :)

Re:Avoid the Netgear SC101

[Anonymous Coward]

Netgear's short-sighted decision to use some strange proprietary protocol instead of SMB

This was funny as hell!

In Soviet Russia standards open YOU!

Re:RAID-5? BAARF!

[cloudmaster]

Write performance: insigificant. He said it was for archival use, so presumably it's a lot of reading and not so much writing. Besides, any reasonable RAID should be faster than a single disk, and with just two or three drives you'll be fast approaching the upper limit of gigabit ethernet (I'm presuming that Taco's house isn't wired with infiniband, though I suppose it might be).

Multi-disk failure: Well, you can still lose your RAID-10 if two disks from the same linear array fail, so you're spending a lot of money and not really gaining a whole lot - the 33.3% figure only applies to a 4-disk RAID-10.

If you've got 4 disks and are concerned about 2 failing, go RAID-6. You get the same capacity as the RAID-10 would get you (capacity * (n-2)), and you also have a 0% chance that 2 failed disks will take the array down. To increase capacity, you just need to add one disk at a time, too (after the initial 3), as opposed to the RAID-10, where you have to add in [at least] pairs.

Re:It's Time my Son

[MikeFM]

Have you tried SFU? It's really crappy. Troublesome to setup and use with major reliability problems. It's workable in a pinch but clearly designed to discourage using it. It just says "We have this feature but look how crappy it is compared to a pure Windows platform." all over it.

Re:The Poor Man's RAID Array

[thogard]

Why is this so hard?
Problem: Disks Die
Problem 2: They aren't free
Problem 3: I can't lose any data at all ever even the stuff from the last 24 hrs.
Oh wait, home users and 99% of business don't have problem 3 even if they think they do.

So you get your self a nice cheap low power box with a big disk(s). Since its on 24x7 maybe even consider some some of the 2.5 inch disks or if you need speed (which you don't) get the 10k 2.5inch disks.
So you get a few external cases with firewire and/or UBS2. Set up cron job based on the speed of the machine and how much data your willing to sacrifice to rsync the external disk to the internal one. Then every so often you can pull the external disk off and swap it with one you keep at the office.
This protects you from any single disk failure and some cases of two disk failure (i.e. lighting strike at the wrong time). You also get an offsite backup and you have a backup of a few days or weeks ago.
Another thing to keep in mind is keep the formatting simple. It means less issues when things break and things will break.

Solution

[MikShapi]

I don't know about all you wackos with the 600$ CPU's and 600$ RAID controllers at home, but I have better things to do with my money. Like invest it rather than spend it on useless trinkets, theoretical seek-time figures and unused gigaflops.

Here's my brew:

1. Old PC. Any one would do, probbably even a good'ol P1. 128MB RAM is more than enough. I consider this FREE. I run a dual-PIII-450MHz that I have lying around.
2. 4x[BIG-SATA-DRIVE]. How big? When I built mine, highest bang-for-buck was 250GB. So I went with 4 of those.
3. 1x PCI SATA controller.
4. 1x PCI GbE NIC.

[3] and [4] are peanuts. [2] is worth, what, 500$?

The entire rig will easily give you ~10-25MB/sec, which is, for any home use I can consider including pumping 10GB files over the network, plain enough.

Plug any crap old 2GB or greater IDE harddrive in for sport (or two and do yourself a RAID1 configuration).
Install Linux.
Install SAMBA.
Configure RAID.
Set up healthchecks that email you if something in /proc/mdstat is wrong.

[OPTIONAL]
1. Grab several old IDE drives. Not neccesarily same sizes.
2. Stick them in some other box (I did it on my windoze box cuz that's where I had case space).
3. Configure a RAID0, or better yet, a spanned volume. Use windoze dynamic disks, use LVM, whatever makes your boat float. Set up a compressed filesystem if you think that would help any. Usually, with the kind of things people store on huge arrays at home, it won't.
4. Do a daily dump of everything from your RAID to your backup array.

DONE. Forget about it and go do something better with your time.