What Kind Of Software RAID Are You Running?
ErikZ asks: "Lately, I'm having issues with my RAID. Specifically, closed source drivers for my RAID card that only support Red Hat 9. So I've decided to Ebay the card, and try to figure out how to turn 4 SATA drives into a software driven RAID 5 setup. Yes, I know I'll lose all the data, and I'm not worried about it. Finding a 4 port (or more) SATA controller card, that's well supported under Linux, has been difficult. Everyone wants to slap on their own RAID chip and charge you another 100$ for the pleasure. Where can a guy get a highly recommended, well supported, 4 port SATA card for Linux? The Rocket 1540 cards have vanished off the face of the earth. There are a few motherboards out there that have 4+ SATA connectors on them, but they also add RAID and some other cutting edge features that aren't well supported under Linux.
So, I thought I'd try another route and ask Slashdot: What are you using for your Linux software RAID needs? What do you suggest?"
Why not just let linux handle the raiding of the drives? No special hardware needed ourside of the drive controllers you already need to hook the drives up.
For more information check: man md
Also RAID 5 is distributed parity raid, no data loss if only one drive goes. it takes two failures to lose data on a raid 5 array.
Software raid? Only on non-production boxes. Otherwise I use some ole' AMI MegaRaid controllers. No software raid for me thank you very much.
What experience I've had with it under Linux systems makes me flee in terror.
It's been running my raid for a couple of years...I think. I cannot honestly remember when I first configured it.
There are some things to be aware of. If you want to mount / as a raid, it can be tricky. The initrd needs to be properly configured, or the drivers must be built into the kernel.
Sometimes, the raids don't shut-down completely. I've never been able to completely solve this problem. Most of the time it's OK, but some machines have trouble. The most common culprit has been NFS.
GRUB & LILO need some special care when you plan to mount root from a raid mirror. It does work and it has all of the appropriate relibility features.
First of all, don't use software raid when you can get, e.g., a nice 3ware card that works well.
But, sometimes software raid is fun to play with at home. For that purpose I just have a couple Promise cards stuck in my machine, each with two SATA plugs. Works fine.
If you're using software raid, it seems to me it doesn't really matter what kind of hardware you are using.
I like this one: MegaRAID SATA 150-4. Admittedly, I've only used it under OS X Server, as it's apparently what Apple uses in their OEM; but they do have linux drivers and I can only assume that they work as well, if not better. Straightforward setup on the CLI, and not too expensive.
Personally, for $300 I wouldn't screw around with a software raid unless this is your own personal box and the drives only have MP3s.
--
$tar -xvf
But can't you just use your raid card as a SATA card, and ignore the raid functionality? Why do you absolutely need it to be non-RAID? I'm sorry, but I'm having real trouble understanding what's the difficulty here...
I've got a Promise TX4 (4 port). I use it under windows for software raid 5, but it has been good to me.
It lists Red hat and Suse support, I thought that it was one of the cards with wide spread linux support though.
I am using the Promise Ultra133TX4 combined with raidtools. Granted this is a 4 port (8 devices) PATA controller but I think the 4 port SATA board they have would suit you fine. Setting up software raid under Gentoo was a breeze. I have 6 160gig Hitachi drives in a raid5 configuration and it has been rock stable for over a year (only shutdown for the huricanes in Florida and a kernel upgrade).
-- http://anonet.org -- The internet the way it was meant to be. Check it out, you may be surprised.
This is not software raid it is a SATA-RAID card so I am not exactly answering your question. Bet this card has worked well for me and they perform well too.
/dev/sda /dev/sda:
# hdparm -tT
Timing cached reads: 2052 MB in 2.00 seconds = 1026.16 MB/sec
Timing buffered disk reads: 380 MB in 3.01 seconds = 126.27 MB/sec
personally?
i'de get this board, not worry about the SLI features (unless your gonna use it) and not care about the hardware RAID... this way if you get really paranoid you can stick 4 more drives in and mirror the bitches! or alturnativly stick in more drives to make it go faster!
but thats just what i'de do. if it was just for me servin files to myself & friends. anything more i'de start to look into more $$$
---- The first point-and-click interface was a Smith & Wesson
I raid whoever I like.
Sincerely, BSA
The best option for real hardware SATA (or IDE) RAID in Linux is 3ware, bar none. Their drivers have been in the official Linux kernel since the early 2.4 days, and they just work. Highly recommended.
Why real hardware RAID? Say, for example, your boot drive goes out in a software RAID configuration. Your system is suddenly unable to boot, requiring manual intervention for a rebuild. With hardware RAID, the BIOS built-in to the card handles things smoothly and your system can boot without a problem.
I've had major problems with Promise cards under Windows XP. Promise RAID is part software, part hardware. RocketRAIDs have had the same problems. There seems to be basic problems, like Microsoft may not want other RAIDs competing with their Windows 2003 software RAID. Windows seems to have timing problems that confuse RAID cards.
The problem seems to be detecting when the RAID array is broken. This problem has gotten much worse with faster motherboards, because the timing window is shorter. If so, then Linux would be affected, also.
In any case, under Windows, the Promise and RocketRAID cards would often detect a broken mirror or RAID when it was not actually broken.
These problems occur apparently only when booting from the RAID array. When using RAID only for data, there seems to be no problem.
You know you've seen one too many memes on LiveJournal and other such places when you just read that "What Kind of Software RAID Are You?" and think somebody's posted a link to a quiz.
Warning: Apple/Nintendo fangirl. Likes her electronics cute & cuddly. May be rabid.
Highpoint makes great cards!
check out their controllers and they work under every OS, well at that!
Of all the PC RAID card manufacturers, 3Ware has the best reputation. However, you cannot boot from one drive in a 2-drive mirror. If for some reason you don't have a working 3Ware card, you cannot get your data. It is lost.
If you use 3Ware cards, keep one or two spare cards.
HighPoint RocketRAID cards do not function well when used as the boot device for Windows XP. This was verified by HighPoint technical support. We did not try them under Linux. But read my comment above about timing issues.
5x Maxtor 6 B200M0 sata 2 -- ~$120 ... -- def not priceless, but it has pcix and is spare *shrug
/dev/md0. formatted reiserfs 3.6.
in a supermicro 5 slot sata hotplug raid enclosure. -- ~$100
connected to a highpoint rocketraid 1820a 64/133 pcix 8-slot sata card. -- ~$150
2x xeon 2.0Ghz @ 533 fsb
integrated intel 1000MT gige controller on mb
500? watt ssi ps, forget the brand... -- ~$75
gentoo linux, kernel 2.6.9, highpoint card has a proprietary driver which sucks, and proprietary software raid module which gave me problems i'll explain later but would probably be workable if you are into that kind of thing.
linux md raid-5 across all 5. 800Gb resultant
my opinion:
works well enough, performance is not amazing but is very solid under most/all loads (mostly fileserving on nfs/samba)
my biggest problem is that the hotplug enclosure tends to cause at least 1 drive to fail to start up properly on reboot, needing a removal and reinsertion and rmmod/insmod to get all the drives online, which is a royal pita. This also means I can't use the built-in software raid module, which doesn't bother me that much, except I lose the hotplugging (not something I'd use). Once I'm up and running tho it's rock-solid for as long as I need it.
md has been beautiful to me so far, too bad the drives support ncq and the card does not. Highly recommend something similar, unless you can get a better controller. I read lots of reviews and few hw sata raid cards came close to sw raid in linux, and the flexibility of md is a MUST-have. Don't think I'm hitting any bottlenecks yet. Really hard to beat the storage/price point right now, and at that size backups become non-trivial.
note: samba and nfs clients will probably require mild to severe tuning for best performance, ymmv. For mysql/apache loads this is possibly overkill, but md is very good with cache/paging if you have the ram for it. Most/all other linux hw raid controllers get killed on cache management alone from what ive read.
Also have a 4x 6y250m0 250Gb raid for backup purposes, but that is usually offline and uses a secondary raid controller.
pricing is half-guess/half-bs
The first rule of USENET is you do not talk about USENET.
Why not a rocketraid 1640? They support 4 SATA drives and support (so they claim) Linux. I've run a Highpoint card under FreeBSD with no problems whatsoever...well, the management software won't work, but, hey, I can check things with the command line....
-- Fugacity: Confusing chemists since 1908
A righteous and just one against those godless peer-to-peer commies, thank you very much.
I took some notes when I was setting up my home 4x250GB RAID 5 server. I found there to be three categories of RAID solutions. Might be helpful for you in deciding. Copied them below and added a few extra comments.
t ml#ss4.1
It really depends on what you are using the server for. I ended up going for the pure Software RAID option. Its for home and I'm a cheap. If you're not cheap or it is for a work server, I'd stick with the pure Hardware solutions.
________
Hardware RAID:
The expensive Adaptec, 3ware, etc SCSI cards found in most servers.
Pro - Offloads XOR calculations from the CPU to internal processor.
Pro - No manual intervention required in case of a raid failure.
Con - Expensive.
Con - Third Party and/or closed source drivers often required.
Semi-Hardware RAID:
These are typically the SATA RAID controllers that are built into motherboards but includes cheapo add-on cards. Generally if less than 150 bucks, not full hardware RAID. I believe all of the RocketRAID cards fall in this section.
Pro - No manual intervention in case of a disk failure.
Pro - Cheap.
Con - Minimal or No CPU offloading.
Con - Third Party and/or closed source drivers often required.
Software RAID:
Use Linux and plain old SATA/PATA controllers to handle all of your RAID needs.
Pro - Very cheap.
Pro - No worry about driver incompatiblity or closed source drivers.
Con - No CPU offloading. You essentially trade CPU power for Disk speed/redundancy... and its a significant trade.
Con - Manual intervetion required in case of disk failure.
Con - PATA Only. Must be one drive per channel! NO SLAVES! Apparently data loss can occur on both drives in the chain if one goes bad. http://www.tldp.org/HOWTO/Software-RAID-HOWTO-4.h
Performance is also hurt in a Master/Slave combo.
________
The general consensus among linux kernel engineers and software RAID users is:
1. As long as the onboard SATA chip is well supported on your linux kernel, use the onboard chip.
2. Don't worry about the "hardware RAID" built into the motherboard. You don't have to use it. In fact, most people bypass it.
3. Use the non-BIOS SATA driver for your motherboard. Some motherboards have two different chips. Mine (an Epox 8RDA+Pro nForce Ultra2/400) uses both the common Silicon Image SIL3114 which supports 4 SATA drives and an additional 2 SATA drives provided by the onboard nForce 2 Ultra Gigabit MCP chipset. Quite nice for RAID and I still have normal PATA IDE drives 0 - 3.
4. Quite often the SATA RAID hardware only supports RAID 0,1 and 10 (or 01 depending). If you're looking for RAID 5 then you'll have to buy a more expensive outboard solution. The problem with outboard solutions are that they will eat into your PCI bandwidth. If you will be using PCI-X then you will probably also be paying significantly more for your outboard solution. Most people have a ton of CPU lying around, so handing off the I/O doesn't really buy you that much.
5. When it comes down to it you might as well just use software RAID because you have more control over it. You can use the onboard SATA controllers which allow you to take advantage of the increased on-motherboard bandwidth as well as having a significantly less expensive solution.
6. Another advantage to using Linux software RAID is that you don't have to learn a new RAID management system everytime you upgrade your machine and controller. You can also connect to your machine remotely and manage your raid system through a firewall. Sometimes you can do that with your hardware RAID system and sometimes you need to manage it from the BIOS itself.
7. Once you get comfortable with software RAID you can experiment with mixing and matching various I/O systems underneath it. One of the things I'd like to play with would be using software RAID with Firewire 800 external drives in a pseudo-SCSI arrangement.
8. The LVM2 system doesn't need software RAID, but it works very nicely with it none-the-less and gives you snapshot support etc.
9. Personally, I'm going for RAID 10 (striped mirroring) because drives have gotten very inexpensive and I don't mind burning a few more to get higher I/O rates. Remember, if you go with a mixture of RAID 0 and 1 then you want a striping over mirroring -- that way if you have a single drive failure the array keeps going.
Have fun and don't use RAID instead of backups. Backups save the stuff that you deleted intentionally but need to recover.
If I might add a comment to your last point. IDE drives take a performance hit when you are working with master and slave devices period. This is not a tradeoff of software raid but just a general issue with ide. You can run software raid across master and slave devices just fine, I do right now both at home and on 168 linux boxes where I work, and we hold the service those boxes provide to 5 9's. The only issue I have ever had with software raid has been throughput. You just cant beat a nice fiberchannel enclosure and a massive controller card to get the data rates up, but that is why we put our databases on the SAN and use software raid for the frontends. Now, yes the howto says not to have both of your drives on the same ide channel in case the bus itself freaks out and starts puking bits all over your drives, but how many people have you met or heard of that have lost data because their ide bus died? Not saying it can't happen, just saying that for most of us, this really isn't a huge concern. Besides, we all have backups, right?
www.linux-skunkworks.com
Thanks for the great post.
I would add to software RAID:
Pro - More inspectable than a cheap RAID card
My $0.02
I've setup a few small office boxes with RAID1 - the average CPU today has power to spare serving RAID1 over 100MB connection - most can do a 1G connection too. Save hardware RAID for big time servers or the gamers.
.\.\att Clare
I don't usually like responding to my own posts, but I forgot to add something.
Whatever software RAID setup you choose, install the test failure driver in the kernel. That way you can force a simulated failure and make sure the system takes a licking and keeps on ticking.
Good luck...
Software RAID: ...
t ml#ss4.1
Use Linux and plain old SATA/PATA controllers to handle all of your RAID needs.
Con - PATA Only. Must be one drive per channel! NO SLAVES! Apparently data loss can occur on both drives in the chain if one goes bad. http://www.tldp.org/HOWTO/Software-RAID-HOWTO-4.h
Performance is also hurt in a Master/Slave combo.
Um...why PATA only? I've done software raid using loopback devices. I can't see why SATA drives would be any less likely to work.
tasks(723) drafts(105) languages(484) examples(29106)
I learned about the Promise SATA150 TX4 from the aaltonen forums. The card is just a disk controller and support is in the 2.6 kernel. I'm using it in a software raid5 configuration and haven't had any problems. It's about $75 at newegg.
GET YOUR WEAPONS READY! --DR.LIGHT
Nothing fancy... just a RAID-1 set, using Linux md.
Since a desktop/workstation machine does mostly reads anyway, I am getting the benefit of striped reads. I don't really care that my writes incur a slight penalty.
Granted, hardware RAID would use less CPU time... but hardware RAID is tied to a particular card. What happens if you move your disks to a new machine? You have to move the RAID card. If you go with an integrated RAID solution on your motherboard, that's tough.
With Linux md RAID, that is not a problem. Just plug the drives into your new machine and go.
Be seeing you.
scott
I've considered setting up software raid on my Linux server, but I haven't found any doc yet about what happens in the event of an unexpected crash or poweroff part way through writing a RAID-5 stripe.
Suppose I have 4+1 disks in a RAID-5 configuration, and during a write to a stripe of the disk only two of the disks are written to before the system crashes. This leaves me with 2 disks with new content, 2 disks with old content, and a useless parity.
I found a page at RedHat that indicates that as of 2001 there was no multistage commit. Has anything changed since then? Do the Linux MD tools address this?
I like striping on my desktop. If one of my drives decides to fail, I'm hoping to get plenty of warning, and besides, I have the essential stuff backed up to gmailfs.
I'm doing this because the mirroring is a performance AND a disk space hit, and is only worth it if I am planning for the disk to fail. With striping, I lose $20 for buying two 80 gig drives instead of one 160 gig drive, and I get twice the speed.
The annoying part is that I have to redo stuff if I want to add to the array. That's the one advantage of mirroring -- you can hotswap drives and add them to the array, no big deal, whereas with striping, you have to backup/restore. But if I don't have to restore once in awhile, I'll never get my backups right.
Don't thank God, thank a doctor!
Sorry that was just a confusing line. In my original post change "PATA Only" to "This Con only applies to PATA drives". Then it should make more sense.
Read his post - He's trying to find drive controllers that support 4+ SATA drives but *do not* include hardware raid.
He's not asking how to set up the software, he's asking about hardware that contains the features he needs for software RAID (many ports) but not redundant features that reduce compatibility and/or add significant cost. (hardware/firmware RAID).
retrorocket.o not found, launch anyway?
a) Cost
b) There's a good chance that even the non-RAID capabilities of the controller have been compromised by lack of documentation to write a good driver
retrorocket.o not found, launch anyway?
http://www.newegg.com/app/ViewProductDesc.asp?desc ription=15-124-020&depa=0
Syba 4 port cart based on the SiI3114 - $30. Supported by libata... I'm using one right now in my big storage server.
Although I have never set it up myself so it is possible I'm talking out of my ass.
t ml#ss5.9/
From what I read about the Persistent Super-Block and from what The Software-RAID HowTo says, you can boot from a software RAID setup.
http://www.tldp.org/HOWTO/Software-RAID-HOWTO-5.h
That way, you won't have to deal with the lame support and archaic capabilities of the steaming pile that is the LinSux codebase. Not to mention, you'll be more secure and enjoy more stability under Windows than you would with LinSux. LinSux is for zealot losers with a sweatly, precarious grip on reality.
> Software RAID:
/boot is raid1 across 5 disks, the rest is raid5. Read performance for a 5 disk raid1 would probably be fantastic :)
/boot is still readable.
> Con - Manual intervetion required in case of disk failure.
You can get around this for some failure modes, as long as your boot partition is always raid1. I do this at home,
The success of this depends on the disk either failing so badly that the system can't see it anymore and so boots off another disk, or that the part of the failed disk that holds
OK, I'd like to piggyback on this question.
/dev/ataraid/d0pN). But Linux 2.6 can't see it. I've done some reading and from what I can tell it doesn't support the /dev/ataraid tree anymore.
I have a Promise FastTrak100 Lite controller built into my MB, and I've been using it for firmware RAID for about three years now. It worked fine in Windows (using the Promise SCSI emulation drivers) and in Linux 2.4 (via
Is there any way to get a 2.6 kernel to see the array while leaving the data intact?
It's always a long day... 86400 doesn't fit into a short.
i don't know what kind of kernel you run, but my 2.6.11 supports md with SATA and PATA mixted together just fine:
...
Personalities : [raid0] [raid1] [raid5] [multipath]
md0 : active raid1 sda1[1] hdc1[0]
120060736 blocks [2/2] [UU]
sda is a sata, hdc is a pata drive
Never said they couldn't be mixed. You're probably referring to the last line about the PATA problems. See my other reply to Short Circuit. Basically that part of my notes just reads confusing. The notes were meant for my reference. Didn't really write them for others to read. That last Con only applies to PATA configurations. I'm not saying that you cant mix them, nor am I saying that Soft-RAID only works with PATA. IF you use PATA, you must be aware of that Con.
HighPoint RocketRAID cards do not function well when used as the boot device for Windows XP. This was verified by HighPoint technical support. We did not try them under Linux. But read my comment above about timing issues.
/boot partition). My Athlon TB1333C with 512MB Ram has no further problems with generating the parity for Raid5.
I have a Abit KT7A-Raid board with onboard HPT370 Controller. Attached to this one are 3 Seagate 80GB and one Maxtor 20GB drive. The Seagates are combined via Linux Software Raid5 (using Raid1 for the
Otherwise, Windows seems to have serious problems with the Highpoint-Controller. Quite often, i found the Windows partition on the Maxtor scrambled, so I had to use my backup to regenerate it. Attaching this disk to the onboard VIA-controller does not generate these problems. If it wasn't for the games, I would leave the Windows world entirely.
I never boot off of raid so i've never had this problem I use raid for things that need the disk speed, like multi-video streams and games. Other than that I find booting from raid a danger and a waste of time.
Why should booting off a raid be a waste of time? Takes as long / short as booting off a single disk. And if you've ever lost something (perhaps personally) valuable due to a disk failure, you appreciate any kind of security you can get which includes storing your data on a raid array.
I know DRDB, but that's more for HA pairs and cannot sync drives in background while mounted.
On my systems, I have a software RAID-1 "boot drive."
.)
If one drive in the pair fails, things keep ticking along smoothly. They're really just identical partitions with identical data on different disks.
LILO merrily writes boot code to the array without episode. Meanwhile, the machine's BIOS is happy to boot from disks other than primary-master, all by itself.
I've booted the system after randomly unplugging devices. It works just fine.
Why do all of you 3ware goons think that the world wants to buy hardware which offers no clear advantage over having no hardware at all? (As if I want to add -more- potential points of failure to my systems . .
Kid-proof tablet..
Beware of a company that is so sloppy that it won't go to the Amigo Guest House in Taipei and find a U.S. citizen and native speaker of English to help it with the English on its web site: http://www.highpoint-tech.com/
hardware RAID has these advantages:
/
... your raid sets? I don't know, you'll have to investigate this yourself.
1) offloads operations to the controller, so eats less CPU/IO bandwidth.
2) can have battery backed cache
3) often looks like "just a scsi controller" to linux and the boot loaders, so booting from f.e. a RAID5 set is often easier.
software RAID has these advantages
1) is cheaper
2) CPU time lost makes hardly any difference
3) has well-tested and supported tools to manage your raid setup. (imagine if you could only set up your raid sets by rebooting and entering the raid bios)
4) disk-layout is non-proprietary (controller died? don't have the same brand lying around? manufacturer left the market? no problem!) - so all-around more flexibility.
Look here for properly supported sata disk controllers:
http://linux.yyz.us/sata/
Some of these cards come with BIOS smarts that provides you with software raid which offer you the advantage of point 3) of hardware raid, ie: bios and boot loader support for your raid.
however, this does mean that the on-disk layout has to be recognized in linux, so linux can make sense of it and set up the raid sets properly. In linux 2.4 there were some drivers that did that themselves, however for linux 2.6 there's now a little userspace program that recognizes a whole bunch of on-disk layouts, and sets them up using the device-mapper facility (part of LVM2).
The advantage of this is that you can use the same well-tested and -supported linux drivers mentioned on http://linux.yyz.us/sata/ , but still use the (bios) facilities provided by the hardware. Another advantage is that this program will probably be used by all ATARAID ("mostly-software-raid") devices on linux, so it is, or will be well-tested and -supported in itself.
You can find this program, called DMRAID here:
http://people.redhat.com/~heinzm/sw/dmraid
So if you decide to go the SW-RAID way, think and decide if you want the advantage of dmraid. I haven't tried this myself yet, and the only aspect I'm unsure of is the management aspect of it (like with HW-RAID drivers) - DMRAID doesn't use MDADM, so how can you properly monitor, hot-add,
MDADM itself isn't going away any time soon either, if I understand correctly. (And even if it does, it's probably very likely that they'll make DMRAID understand the MDADM on-disk layout to provide an upgrade path.)
If however you decide to go the HW-RAID way, make sure you get a reliable and reputable manufacturer - with open source drivers (!), preferably with a known on-disk layout, and be prepared to spend money. I've heard a lot about 3ware, but I have no direct experience with them myself, so I can't vouch for them.
Booting is not an issue with software raid. Simply use one cylinder of each drive as a boot partition. Mirror the partitions with RAID-1. Booting is a trivial problem.
I'm booting off my RAID 1 drive (mirroring) using Software RAID. For linux, if you want RAID 0 or any other version, I guess you can put the /boot on a NON-RAID partition, create an INITRD that loads the raid modules and set the real-root to the RAID partition and it should work like a charm. So you only have you kernel and modules on the non-RAID device and the rest of the system on the RAID device, so you don't lose the advantages of RAID for system files.
Performance is not noticably slower and I get piece of mind... well, as much as you can do running XP.
Why aren't we discussing the real issue? Should we allow "ebay" to be used as a verb.
Intron: the portion of DNA which expresses nothing useful.
You can't just ask what RAID solution to use; you need to specify your optimization criteria. For example, if you are going for inexpensive, high capacity, and good redundancy but don't care about hotswapping, you can go with a reasonably cheap PATA or SATA hardware RAID. If you are willing to skip high capacity you can get an external SCSI SCA rack, an inexpensive SCSI card, and a bunch of small SCSI SCA drives. If you want high capacity and high performance without too much cost, get a bunch of large PATA drives, a similar number of FireWire or USB 2.0 (or both) enclosures, as many FireWire or USB 2.0 interface cards as necessary, and do it that way (which also gives you hotswap). If you are willing to spend lots and lots of money, get a Network Appliance server.
Personally, I'm not worried about performance beyond being able to play full-motion video. I have a PPC 604 180MHz from 1997 with a SCSI card and a RAID rack. 8x18GB at RAID 5 gives me 118GB or so of redundant storage, and I serve it over NFS to my other machines. Just for kicks, I have it going through a cryptoloop, too (LVM on cryptoloop on Linux RAID5 on SCSI). The initial cost was low (the drives were $15 each, the rack was around $100 on eBay, the trays were given to me, the SCSI card was under $40 on eBay, the 100Mbit ethernet card was about $20, and the computer had been a doorstop until I put Linux on it). The ongoing (electricity and cooling) costs are a little high (they are 10K drives), but that's life. I can play an MPEG or AVI from two machines on the network at once without hiccups, so I'm happy.
If I were going to build a RAID server today, I'd probably buy a Mac Mini, four large PATA drives, and four FireWire enclosures. Assuming 160GB drives, I'd have 320GB of RAID5 storage available over NFS (with a spare drive to swap in) for an investment of under $1200, and I can vary that cost with the size and number of drives. Yes, I'd be daisy-chaining FireWire, which means that each drive has only a portion of the total bandwidth. Then again, my network card will only manage 100MBit, so 3/4 of the FireWire bandwidth will be of minimal use anyway (except for reducing latency due to readahead and such, of course).
I'm not the only one that tested, and concluded, that software RAID outperformed a 3Ware card.
The problem you run into is bandwith.
Picture four drives on a single PCI controller running in a software RAID5. For every block written, four commands must be issued to the PCI card, one for each drive.
This works great w/ a hardware RAID controller, because they emulate a single SCSI drive, thus only one write command.
Even so: we saw better throughput via software raid ( tested via Bonnie++ on a knoppix 'toram' boot )
BTW: I tested every filesystem format with various block sizes that knoppix offers. As was reported earlier, JFS showed the most impressive jack-of-all-trades performance. Too bad I later discovered you only get two options with RedHat Enterprise WS: ext2 or ext3.
Before I part with'em: two pennies weigh ~4.996+/-0.014g, have a zinc core, and the face of Lincoln. You can keep 'em.
The link in your sig is broken.
went over your head didht it ya fucking moron.
hes saying the card doesnt have any onboard raid - theyve just implemented software raid in the drivers for the card to make ot look like hardware raid.
dunno if its true, but that's what he's saying.
you numshit.
I had no problems booting a trial of Windows 2003 server on my server back in 2003. Not only did the server use a RocketRAID 404 card with the recently released RAID 5 firmware, it used an VIA EPIA motherboard.
:)
It was a little pokey, but it was rock solid. I booted off the RAID 5 array.
I also did not have problems. Some motherboards would work correctly for weeks. Then there would be an unexplained failure of the mirror. HighPoint tech. support said they were not able to understand why the failures were occurring. (Promise told me the same thing.) Highpoint said that the failure was common.
I have a fairly extensive background with raid setups but not within Linux nor on an X86 platform so if someone would fill in some blanks, I would be very appreciative.
/boot is mirrored across disk 0 and disk 2 - partition 2
/home is mirrored across disk 1 and disk 3 - partition 1
/boot when convenient. Can anyone confirm that this is what actually happens?
/etc/fstab refer to "md" devices or "disk labels"? If either device names or labels can be used, is there a recommendation for one over the other?
1) The original poster is looking to setup a 4 drive array, RAID-5 prefered and is looking for a 4 port SATA adapter. My recommendation would be get two adapters with two drives each to provide greater redundancy. I am guessing this can be done with stock PCI SATA controllers. Would a configuration such as this have a negative performance impact based on the characteristics of the PCI bus?
2) Assuming a straight software RAID config, what happens during the boot process if a drive is lost? For example, let's assume the following RAID-1 configuration:
Controller 1:
Disk 0
Disk 1
Controller 2:
Disk 2
Disk 3
/ is mirrored across disk 0 and disk 2 - partition 1
No swap - too confusing for this discussion.
Let's say disk 0 on controller 1 fails. I would hope the system will generate some type of message/alert and then continue functioning normally until it could be shutdown to replace the failed drive (too cheap to go hot-swap). Eventually, I get a new disk and shut the system down to replace the failed disk. I reboot and something happens. Ideally I should see the system come up normally with the possible exception of a warning message and then I can remirror / and
3) Based on the configuration described above, should
I would be quite thankfull for any information regarding these question. Thanks.
If VISTA is the answer, you didn't understand the question
For cheap setups, I go with 3ware. For more expensive ones, we use an external raid array with a scsi uplink to the computer. The cache, battery backup, and simplicity of host communication are important advantages to consider. Software raid, while fast and, just scares me.
Maybe if you're using a 486.
The CPU overhead of software RAID 5 is insignificant on any remotely modern machine. Even "ancient" ca. 500Mhz P3s have checksumming speeds over 1GB/sec
http://www.broadcom.com/products/Enterprise-Small+ Office/Storage+Solutions
It's semi software but it has several advantages:
Linux support (I think the driver is OSS or slated to be)
Transform from RAID 1 to RAID 5 (or 10 or 50) on the fly
Can support boot drive RAID nicely
Hot swap
One thing to note: slower writes, faster reads.
And one thing I've been wondering about, obviously you would keep the drives on different IDE channels if possible (hda, and hdc here).
If you also have non-RAID drives on hdb, and a CD-ROM on hdd... that should not overly influence the data speeds of the RAID drives except when there is actual data transfer on hdc/hdd, or would a machine automatically split the available pipe upon having two IDE devices (master/slave) on a given IDE channel?
The Sil3114 on my Asus K8N-E has the 4 sata ports and will do a raid 5.
On Usenet I posted a detailed description of how I built a 2.8TB RAID storage array for under $4100.
The md0_raid5 daemon takes something like 4-9% of the CPU during the write tests, 0% during reads, and 1-2% during the file create/delete tests.
Man this changes my whole view of software raid. I'd give you a +1 Informative if I had a spare point. How bout just a thank you instead?
My bonnie++ results are below if interested (mangled by slashdot):
"bonnie++ -s 1024"
Size 1G
files 16CPU: 2.6GHz Celeron D
RAM: 512MB DDR400
Drives: 4x250GB 7200rpm 8mb PATA Maxtor
FS: Raid5 512k chunk, XFS
Those high %CPU marks from bonnie threw me off. Guess I need to update my notes a bit.
At work we've been testing RAID 0 (strip) setups for performance on a number cruncher machine. Hardware raid isn't always the best.
Our first setup was 2x15kRPM U320 SCSI drives on an LSI MegaRAID controller. Apparently the 2.6 kernel driver has serious issues, because we can't get read performance over 50MB/s. This is slower than reads off a single drive on a vanilla SCSI controller.
Our second setup was the same two drives on an LSI U320 SCSI HBA. The HBA has a 'simple' stripped raid via firmware. This worked well for two drives, we got 140MB/s reads. This is about twice the single drive speed. We added a third drive to this setup and it must be swamping the firmware, because the read speeds went _down_ to 110MB/s with three drives. But when we switch to linux md software raid we got 215MB/s. This told us that we weren't yet swamping the scsi channel or PCI-X bus.
We're hoping to be able to software raid 4 15K rpm drives on a single scsi channel. This should put us around 280-300MB/s. If this works we'll try a 2 channel U320 controller with 4 drives on each channel. By offloading the raid operations from the HBA to the CPU we should be able avoid swamping the HBA processor. The only other issue to consider is weather or not we'll swap the PCI-X133 bus. If this happends well use 2 single channel controllers on seperate PCI-X133 buses. Our goal is to get over 500MB/s read on a software raid array. Really, this is an insane read speed. It is over a quarter of memory speed on these systems. We're doing this on a Dual Opteron 250.
"You never know when some crazed rodent with cold feet might be running loose in your pants."
-Calvin
"Con - PATA Only. Must be one drive per channel!"
Woah! Wait, what?
SATA is one driver per channel.
Democrats or Republicans. They are both taking us to the same place and they are not afraid of us anymore.