Slashdot Mirror
Ideas for a Home Grown Network Attached Storage?
Ken asks: "It seems that consumer level
1TB+ NAS boxes
are all the rage
right now. Being a digital packrat, with several computers/entertainment devices on my home network, I am becoming more interested in getting one of these for my home. Unwilling to dish out 1K or more up front, and possessing a little of the DIY spirit, I would like to build my own NAS and am interested in hardware/software ideas. While the small form factor PC cases are attractive, my NAS will dwell in the basement so I am thinking of a cheap/roomy ATX case with lots of power. I think that integrated gigabit Ethernet capabilities and PCI-Express on the motherboard are a must, as well as Serial ATA HDDs, but what processor/RAM? How strong does a computer really need to be to
serve files? What about the OS? Win2K3 server edition? WinXP Pro? Linux?"
"I have been using Red Hat and then Fedora Core since it came out but only in a workstation role, and I have little experience with other flavors. What file system should I use for maximum compatibility? I will need it to work with Windows, Linux and several UPnP devices. I am planning on starting out with two or three HDDs in a RAID 5 config. and I would like to be able to add more HDDs as space is needed without any major changes. Thanks for any ideas."
If you think you can beat a device like the Buffalo TeraStation go for it, you will be rich! It was shown at CES, and goes on sale next month in the USA for $999. Gigabit Ethernet, 4 250GB hard drives (RAID 0, 1 or 5 support), 4 USB ports to attach additional external storage devices, built in print server for sharing a USB printer, blah blah blah. I'm going to buy 2 of them!
Urge to post... fading... fading... RISING!... fading... fading... gone.
If you're not worried about having it all in one big partition, do what I did. Get a big case that can hold lots of drives, and just keep adding in SATA or IDE expansion cards and drives. It's worked well so far.
If you do want it all on one big raid5 partition, good luck finding a way to add additional disks into it without rebuilding.
Samba.
"The world only exists in your eyes. You can make it as big or as small as you want." - F Scott Fitzgerald
Common linux file systems (ext, reiser, etc) contains critical data-losing type bugs on file systems bigger than 2TB, except XFS. This was found to be the case in even the most recent 2.6 kernels.
Tony Battersby posted a patch to the LBD mailing list recently to address the ones he could find, but lacking a full audit, you probably shouldn't use any filesystem other than XFS.
Considering the gravity of these bugs, you might consider using XFS for everything, if the developers left these critical bugs in for so long, it makes you wonder about the general quality of the filesystems.
I've had enough abrasive sigs. Kittens are cute and fuzzy.
I feel strange advocating a MS-originated protocol -- but the truth us, serving files via Samba on Linux is going to be the best-performing[1], most-compatible remote file system available.
As for hardware, for small servers I like Linux software RAID, but for a big multidisk farm, you can't beat 3Ware cards. They take nice cheap IDE drives and turn them into a SCSI RAID. Moderately expensive, but beautifully functional. Finally, I've been having good luck with Seagate and WD drives, and bad luck with Maxtors. Your mileage may vary.
[1] Samba beats the MS implementations of SMB/CIFS. No guarantees about Samba vs NFS, GFS, Coda, whatever.
I was thinking of using a mini and a single firewire disk for a somewhat similar project.
But, OS X has RAID capability, so you could use something like this:
Lemme get this straight. You asked Slashdot whether you should use Linux or Windows? Do you never read the comments around here?
;)
Oh, wait.
I suppose this is Slashdot. Nevermind.
That green slime had it coming.
Buy everything piecemeal. I just priced out a 900Gb NAS for $800, shipping included. Slap it all together, put your favorite Linux distro on it, and run Samba.
You won't be able to beat the price of the real thing by much, though: big hard drives are still expensive, and so are RAID cards (if you go that route).
Another one bites the dust
My current fileserver is an aging 600MHz P3, with 4 PCI slots, each occupied by a raid card. I've decked it out with 250GB disks, 4 on each card, 2 on the mobo, and 2 100GB disks on the mobo for swap,boot,and root. I've got onboard 100Mb ethernet, no graphics card, no sound. I've installed gentoo with a stripped down kernel, running samba, and it all works beautifully.
It's a fairly large box, in a full size ATX case, and the disks are also stored in a rack which I built and bolted onto the side of the case, as I ran out of space. Each partition is 1TB, using raid-0 (linear), as I don't care about redundancy too much. Power is supplied by a 450W supply, and cooling by a stack of old 6" server fans. Also, it lives in an outside closet, which helps to keep it cool.
In all, the setup has been relatively cheap, all the hardware is from eBay, and I've yet to have a problem.
I'm also running a bootp server off it, and am netbooting two media stations from it!
I would also like to build such a thing, but a box full of disks spinning 24/7 is likely to use a lot of power and give off a lot of heat. Are there any power saving solutions to this? It would be nice if there was some intelligent software that, when you try to play a movie off the disk, spin up only the disk that has the file, read a large chunk of it into memory, and spin the disk back down.
Is this doable?
When you're dealing with that much storage, you really need to catagorize your files into what needs to be backed up and what doesn't. In this type of application (if it was me), most of the storage is likely to be filled with dvd rips & mythtv recordings, or backups from your main system(s). So you would want to backup a list of what you have, but you can always recover from original media (in the case of dvd rips, or off of re-runs for tv shows). Also, on a storage server you're more likely to have data loss from physical driver failure (hence the raid 1 or 5), Since you won't be playing with the system that much once it's set up, you remove a lot of risk factors that you'd have on a desktop system (accidental file deletion, filesystem corruption, ...)
Imagine this with a high-performance SATA raid controller [1] [2], in an enclosure barely bigger than the 4 hard drives alone.
Someone knows here to buy this motherboard? What about practical experience with this sort of configuration?
Wow, a front page article on Slashdot that amounts to, "gee, how do I build a server?" Spiffed up a bit with trendy, techie-sounding words, but cripes. This is FP news-worthy?
That said... if all you're doing is file serving, a tiny machine by modern standards is fine. 64 megs of ram in a P3/400 would make a very solid home server. If you want to use software RAID, though, it's a good idea to go faster... you'd want at least 1ghz for that, maybe 2, depending on how much traffic you were sending to the box and how patient/impatient you are.
Since it's going in your basement and you have no worries about size or noise levels, get a big whompin' case with lots of 5.25" slots. Cremax makes some nice enclosures that will let you put 5 3.5" drives into 3 5.25" bays, with good fans for cooling. They have multiple variants. I'm using the SCSI flavor, but you can get them in SATA too (and IDE, I think, but I'm less sure about that.)
I have an older 3ware 8500 RAID card, and it's dismally slow at RAID 5, even though it's supposedly 'optimized' for it. I don't know if the newer SATA versions are better, but while they are well-supported in Linux, and, being hardware RAID, are a total no-brainer from an admin perspective, my generation of cards was horribly, horribly slow. I get at least four times the performance using Software RAID on an Athlon 1900+.
This is how my network server looks:
Big case;
400W PC Power and Cooling power supply;
ASUS A7V333 motherboard;
Athlon 1900+, I think just 266mhz FSB (not sure);
1 gig of RAM (nice for caching, not at all necessary to have this much);
Ancient video card, Matrox Milllenium 2, I think;
3com 3c509 network card;
ICP Vortex 32-bit RAID controller, bought used. The first one I got was dead... had to replace it. I got it pretty cheap, intending it for another project that fell through, and so I ended up using it at home instead. I think it was about $100, but I'm not sure now. These boards KICK ASS. Great linux support, VERY fast. Awesome hardware.
6 18-gig 10KRPM SCSI drives; machine boots from this array, and Debian is installed here;
2 Cremax 5-in-3 SCSI enclosures;
1 3ware 8500+, in JBOD mode (software RAID is WAY faster);
4 80 gig IDE drives (small, but I set this part of the system up a long time ago)
The SCSI array is damn fast, an excellent spot for interactive, disk-intensive things like IMAP or big compiles, while the slower IDE array is ideal for filesharing.
You should be able to set up a similar system for, oh, $1500? And keep in mind... this is HUGE overkill for a home network, it would be a solid backbone for a company up to about 50 people... though it might need more drive space, and I'd probably want redundant power supplies in a really central machine. You could run mail, internal DNS, DHCP, a squid proxy, internal webserver, and Samba for that many people without it even working that hard.
File sharing is fundamentally a tremendously simple thing, and it just hardly takes anything at all to do a perfectly fine job. Once upon a time this was akin to rocket science, but at this point, even a garbage $200 PC from Walmart would probably be an okay fileserver.
Again: the specs on the machine above are wild overkill... swatting a fly with a sledgehammer. But if you want to spend that much money, or you have most of the parts laying around the house anyway, it'll do a damn good job.
Well, the LBD mailing list archive might be a good place to start, since I specifically mentioned it.
Patch 1
Patch 2
Says Tony:
"Here is an "example" patch to fix some of the LBD issues with various
filesystems (ext3, xfs, reiserfs, afs). Unfortunately it looks like
there are many more LBD problems with the filesystems that I didn't fix,
so I am just calling this an "example" patch that shows some of what
needs to be done, but doesn't fix everything."
He later mentions the only XFS fix is in some debugging code, and it appears to be the cleanest of them.
I've had enough abrasive sigs. Kittens are cute and fuzzy.
A few comments about this
-Get the best-value processor that you can find. You won't need the fastest thing out there, but it's better to have a little more "oomph" than you need. If you end up using an encrypted filesystem at some point, you'll want enough power to decript and keep the network "fed"
-Have a plan for adding a second network interface. Maybe you don't need it now, but once the DIY bug bites, you may find yourself wanting to use the machine as your NAT box or as a wireless access point or something like that.
-Think about noise and power use. Yeah, those WD Raptors are fast, but they're really loud, too, particularly if you buy a pile of them. You might want to think about acoustic material for the inside of the case -- your local car customizing shop can hook you up. You'll also want an "overkill" power supply for the case so that you don't have problems when you add more drives later.
-Think about heat and airflow. At this time of the year, it's easy to ignore (Dear Australia: yes, I know it's summer there now), but during the summer, stuffing the fileserver into the closet might not be such a good idea.
-Consider underclocking. If you do buy a better processor than you need, bump the speed down for now. Less power, less heat, less noise.
-Get a BIOS or hardware-level RAID mirror for your "root" disk. You can use software RAID for the data disks, but you want to be absolutely certain that you can recover the disk with information about the software RAID. The RAID does no good if you don't know how to access it.
-If you use Linux, LVM will become your new best friend.
-Consider buying hard drives that are carried by your nearest Best Buy/CompUSA/other computer store. You don't actually have to buy the initial batch from there, but if a drive in the RAID set goes bad, you'll want to replace it ASAP. It's nice if you can do that tonight rather than "in a few days".
--
When we developed the PetaBox at The Archive, the idea was to use off-the-shelf PC hardware and maximize GB/buck, while keeping cooling and power costs low. It's worked out pretty well. See also my unofficial PetaBox web page.
It turns out that you really don't need much of a PC to serve files. We underclocked the cheap little Via C3 processors to 800MHz to reduce power and heat, and they still troop along nicely. SATA is not necessary, since you're going to be bottlenecked on the network connection anyway. We used 512MB of RAM per node, but only because our system runs a gaggle of perl scripts to provide a variety of services (file searches, XML-based metadata updates, etc). If you're just going to be running NFS or Samba, 256MB is probably plenty (unless you choose to run Gigabit over a mere 32-bit PCI bus, in which case 512MB or 1GB would be better, so that you're reading more from filesystem cache and pounding the hard drives over your overloaded bus less). Gigabit ethernet is a must (we used 100bT for the PetaBox, which is annoying at times, but the cheaper 100bT 48-port switches were instrumental in keeping the overall price of the system low). We stuck four hard drives in each case, mostly from previous bad experiences trying to work with eight-disk machines. I can't say too much about the disk failure rate statistics which incited us to switch to Hitachi Deskstars, but I will say that I'm glad our PetaBox is using Deskstars and I will only use Deskstars in my workstation at home.
If you really, really want to keep the gigabit pipe full while pounding on your disks, then a newer bus like PCI-Express is necessary. Otherwise, I'd be tempted to go with an older, cheaper (and imo, more reliable) Pentium-II or -III based PC. You can get solid, reliable, well-cooled and well-dustfiltered early model VA Linux servers with 500MHz Pentium-III's for $200 or less. I must stress the importance of buying a really solid, rigid case. Over time, normal computer cases get all bendy-wendy, turning every part into a moving part, including parts you don't want to have moving at all. Fans will start sticking, motherboard traces will start breaking, etc. Most of the rack-mountable cases are made of good thick solid steel panels, which makes them heavy as f**kall, but IMO that's a small price to pay for a system that will run forever.
For operating system, the most important thing is to get something you know how to run and maintain, or can get help running and maintaining. If you have geek friends who are willing to provide technical assistance, find out what they know best and use that. A well-known operating system will probably be of more use to you than a technically better, but less well understood, operating system.
Having said that, my personal preference is Slackware Linux, because I appreciate its philosophy of keeping things simple, and preferences for packages which are the most stable, as opposed to newest versions or lots of features. My second choice would be FreeBSD. Third would be the OS we decided to use at The Archive for the PetaBox nodes, Debian Linux. But if all you know is Windows, then go ahead and use Windows.
Regarding RAID, it's been my experience working at The Archive that RAID is often more trouble than it's worth, especially when it comes to data recovery. In theory, recovery is easy, you just replace a bad disk and it will rebuild the missing data, and you're good to go. In practice, though, you will often not notice that one of your disks are borked until two disks or borked (or however many it takes for your RAID system to stop working), and then you have a major pain in the ass on your hands. At least with one filesystem per disk, you can attempt to save the filesystem by dd'ing the entire raw partition contents onto a different physical drive of same make + model, skipping bad sectors, and then running fsck on the good drive. But if you have
Please open your own computer and take a look
before posting misinformation.
I just checked the spec. Like most other
SFF, it has only ONE internal 3.5 storage bay.