MacInTouch Reader Reports

Hard Drives: RAID

Aug. 18, 2009
Aug. 19, 2009
Aug. 20, 2009
Aug. 21, 2009
Aug. 22, 2009
Dec. 30, 2009
Dec. 31, 2009
Jan. 4, 2010
Jan. 5, 2010
Jan. 6, 2010
Jan. 7, 2010
Jan. 8, 2010
Jan. 9, 2010
Jan. 11, 2010
Jan. 12, 2010
Jan. 30, 2010
Feb. 1, 2010
Feb. 6, 2010
Mar. 1, 2010
Mar. 2, 2010
May. 27, 2010
May. 28, 2010
Oct. 2, 2010

Newer entries...
Aug. 18, 2009

item.97791

Allen Huffman

For my data needs, I have been using two dual-bay iStarUSA v7age220 enclosures, and have *ten* 1TB Western Digital Black drives I clone using Carbon Copy Clones (so, five drives of data, backed to a second drive in case of hardware failure or file system corruption). This has worked well, but was cumbersome to swap drives all the time for specific video projects.

I recently purchased a Drobo (current 2nd generation model) thanks to the price after rebate finally going below $300. Since a similar FireWire 4-bay trayless RAID enclousure I had been looking at was around $200, I figured the benefits of Drobo's "BeyondRAID" might be worth the extra expense.

My initial few days with Drobo were nothing sort of magical, though I was frustrated at the lack of polish on some of Data Robotics tools. The website registration form, for instance, failed to accept a purchase date of "7/31/2009" for instace, requiring a leading zero to be added to the month ("07"). Also, even though the registration form was opened *from* the Drobo Dashboard app, it didn't automatically pass in the drive's serial number, requiring you to manually type it in -- and the Dashboard app does not make the serial number Copy/Paste-able. Small things like that made me wonder how robust the device itself would be.

Within a few days of loading up the device (initially with 1TB drives, then slowly swapping in 1.5TB drives to watch it work), I started seeing quirks in the Dashboard where it would not show proper names of volumes. All four of my 1TB volumes would show up as "Drobo1" instead of the specific names I had given them -- so if I clicked Rename, would it trash my existing volume names and set them all to the same? Things like that.

Still, within a week or two of my first Drobo experience, I ordered a second unit -- that way, if one died, I'd at least have a backup to swap drives to and get access to my data immediately. My second device is currently loaded up with four 1TB drives.

A few days after getting about six main drives backed up to the two Drobos, I started seeing more quirks. Sometimes a unit won't show up in Dashboard, or the lights remain on even though I've shut down (I've had to unplug and reset each device once to make it show up again). I also just saw a message claiming a new unformated drive was detected -- this message is expected when you rotate in or add a new drive, but I do not know why I would see it on a unit that has been running for days well-loaded with data. Is a drive dying? Did it get corrupt?

So I thought I'd ask here and see if other Drobo owners are seeing quirks like I am. If these devices were as flaky as mine seem to be, surely so many folks (like Scott Bourne, Geekbrief TV, etc.) wouldn't be relying on them for their impotant data...

Anyone?

Aug. 19, 2009

item.97822

Rob Wyatt

I've commented about Drobo here before. I had a terrible experience with my unit. Twice, when applying Drobo's firmware updates, the volume information on my Drobo was hosed. The first time, the Drobo was able to rebuild itself after four days.

The second time, I wasn't so lucky. I had to purchase a copy of DataRescue II (the *BEST* data recovery app I've ever used...truly magic) to restore my data. Furthermore, Data Robotics help was useless. Despite having a user with total data loss due to THEIR firmware update, they never returned my calls and only promised to have an answer "soon" via email.

After three weeks of waiting for help, I got so ticked off that I started flaming them publicly in every forum and on every review I could find (emailing copies to their customer support and PR people) until they agreed to refund my money and take back the faulty unit. Needless to say, an all-around terrible experience with both the hardware and customer service.

Given Allen's experience, it sounds like not much has changed with Drobo. Bottom line, if you trust these boxes with your data, you're taking a risk. If you really need RAID5, pony up the extra few hundred bucks and buy a proper RAID5 case. One crashed Drobo is days wasted. Surely your time is worth more than that.

item.97834

Dave Howe

Have 4-bay Drobo with DroboShare and see nothing like what Alan Huffman is experiencing...

item.97853

Mark Charnley

I have been using a Drobo Pro with eight one-terabyte drives for about two months and it has worked without any problems so far!

item.97859

Mark Scott

To Allen Huffman:

I have two Drobo units - 1 at home used for a 2+TB itunes library and 1 at work for data. The devices have worked flawlessly for the 6 months I've had them. I have each configured as a *single* 8TB drive. I suspect that your reported issues relate to trying to make the Drobo behave as multiple drives. Also be sure that you have firmly seated the drives. I had 1 drive "disappear" because I had not firmly seated it. If you are using Seagate drives, there are known issues with some firmware versions - check the Drobo site for the offending models. By the way, the Seagate problem is not unique to the Drobo.

The ability to hot-swap drives or to add larger (2 TB) drives and not lose data is amazing.

As with you, my only worry is if the Drobo unit itself (not individual hard drives) fails...

item.97860

Mike Kobb

Allen Huffman asked about experience with Drobo.

I'm currently using a Drobo Pro connected to a Mac Mini as my Time Machine and iTunes repository. I haven't had the specific problems that Allen has had, but I have certainly experienced software quality issues with the Drobo Mac software and their web site. They do seem to suffer from a real QA issue.

I've also had a lot of trouble with FireWire -- if I do large transfers to the unit over FireWire, the unit hangs and drops offline. I switched to iSCSI, and that connection has been trouble-free. The Drobo support folks have already replaced the unit (same problem), and are being very good about investigating, but it has required many hours of my time to set up and execute test scenarios. Since iSCSI is trouble free, I could just use it and ignore FireWire, but I want to get to the bottom of the problem in case I decide I want to use FireWire in the future.

I had considered using a Netgear ReadyNAS Pro Business Edition as my backup and iTunes repository, and I'm sort of wishing that I had. The Drobo Pro had a better form factor (rack mountable) and higher capacity (8 drives, compared to 6), but I think the Netgear's software is much more mature and flexible. For example, it supports multiple volumes of multiple different sizes, and it can support multiple iSCSI LUNs.

item.97881

Michael O'Brien

I have a first-generation Drobo, and it's been fine. On one of my machines, the dashboard app can't make it stand by, but I've found no problems in just unmounting the volumes from the desktop and then pulling the USB cable. That makes the Drobo go to sleep every time. And, frankly, that's the only quirk I've seen. (On the other machine, the dashboard app is always able to put it in standby.)

I've replaced two 1-Tb drives with 2-Tb drives. Each rebuild took about five days, but they all completed fine and I've lost no data. Of course, I haven't yet had a drive failure, either. To be fair, my Drobo spends most of its time powered up but asleep. I only bring it up when I'm actually transferring or using data. That amounts to about 2-3 hours/day.

item.97894

Colin Thompson

I've had a Drobo with 4 x 1.5TB drives connected to my Mac Pro via firewire for several months, using it as my Time Machine - I've had no problems at all so far.

I also have a firewire 800 dual enclosure with 2 x 1TB drives in Raid1 which I use as a second backup + several other external drives where I make more backups of specific folders.

Plus my main internal drive is 2 x 1TB drives in Raid1

You can NEVER be too careful or have too many backups!

Remember what Confucious said - 'Belt+Braces+belt+string+belt mean trousers never fall down.'

item.97908

Brian Lackey

The Drobo needs to have a specific space set during initialization. I started with 2TB but when I added more drives than what would total 2TB, it wanted to make an additional volume with the remaining space. I own 2 Drobos myself and both are set to 16TB even though there are only 3TB of actual space. This leave room for growth. Did you set the initial space to something larger than your total *real* hard drive space? In other words, if you have 3 1TB drives and only 2TBs set during the initialization, it would want to format a new volume containing 1TB.

item.97910

Brian Lackey

I am not impressed with the speed of the Drobo, but I do love that I can have one 4TB drive as seen by the finder. This way I can keep all my video on one virtual drive based off of 4 real drives. This saves from having to save video over multiple drives.

Aug. 20, 2009

item.97936

Brian Timares

Allen Huffman wonders about trusting his data to Dobros.

I do not have experience with Drobos because they are proprietary. There are many similar products out there that use open source software, which provide security in case the company folds, or decides to 'concentrate its resources' on, say, newer products only. I can't imagine trusting my data to a black box where I really don't have recourse other than hope if something goes wrong. "Hope is not a method."

The descriptions I've seen of how they actually store data are long on hype and short on detail. To me this is a warning sign to stay away.

I use Apple's RAID [software, in Disk Utility] because:
1) Hard drives are cheap so mirroring is fine
2) I am OK with managing it (it does need management, albeit not much)
3) I'm too cheap to buy SoftRAID when Apple's RAID works good enough for me.

If SoftRAID had RAID 4 or 5 I'd likely buy it, but I suspect not too many others would. Secretly I've been hoping for ZFS. If you want RAID (you should) and don't want to manage Apple's RAID, I'd say give SoftRAID a good look, and check out the many other hardware products out there that use (typically) Linux under the hood.

Side note: although SoftRAID is propriety, I find them to be more forthcoming technically, and I have alternate reasons to trust them (an expert's recommendation, many years of building up a great reputation, their responsiveness).

item.97954

Larry Nelson

This is more informational, for the curious, than a question.

I bought a Guardian Maximus RAID-1 system from Other World, to use as a backup for my music library. It came with a backup program called Data Backup, so I used that to copy the music library from its home on a LaCie 500GB USB-2 drive to the Guardian. This took about 14 hours for my 460GB library.

I wanted the RAID copy to be the working library, so I went through the steps to have Itunes use that. That proceeded normally -- I've done this before -- but when I tried to play a song, Itunes couldn't find it. If I navigated to the track manually it would play, but I don't really feel like doing that for 23,000 tracks.

I sent a tech support question about this to Prosoft, the makers of the software, and got a prompt response suggesting that I make a "clone" backup instead of the standard type. This makes sense. The manual says that the standard backup places some information on the drive that Data Backup uses for future backups. A clone is just that.

I can't test this at the moment because I don't have a big enough drive. I guess I need to buy another Guardian to use as the working library, and use this one for the regular backup.

item.97958

Shawn Parr

I have a 2nd Gen Drobo (with Firewire) and have had some glitches. I keep mine connected to an Airport Extreme Base Station for Time Machine backups and other manual archives.

The first unit I had was replaced. It worked perfectly for 3 weeks, then started randomly going offline. While connected to the AEBS it should always stay online. When it went offline, the AEBS still would 'see' the volume, and attempt to share it out, but of course no data transfer could occur. To fix it I would have to reboot the Drobo and reboot the AEBS *twice*. The first time I would have to pull the power to reboot it (trying to use Airport Admin would cause it to totally lock up after the drobo unexpectedly went offline), then using Airport Admin to do a software restart.

These offline moments would happen first a week or so apart, but then started getting more frequent, until it was several times a day. While working with a tech from Data Robotics over email and the phone, we eventually got it to go offline when it wasn't supposed to while connected to my Macbook Pro (it took several days before this happened). They then replaced the unit and the power supply.

The offline issue still happens, but it tends to happen once per month or two rather than once per week. And thus far it doesn't seem to be getting more frequent (been in service since March).

Luckily I've had no data loss, even with it going offline while transferring data/Time Machine backups in process. Of course any manual data transfers I was doing while it went offline I would just delete and start over, as it dying while a file was in the middle of transfer is a guaranteed corrupted file. But I haven't had any issues with the volume becoming unreadable.

Every time I've connected it to a machine (usually for short periods, so take with a grain of salt) it has behaved perfectly, especially while using Firewire.

item.97993

MacInTouch Reader

Hi,

Regarding Drobo and the NetGear ReadyNAS:

I have 2 x 4 bay Drobo's (one original USB, one Firewire) and knock on wood, they have been working nicely. (The Firewire has two 2 TB drives and 2 1.5 TB drives, the USB four 2 TB drives).

I also recently got a Drobo Pro (8 bay - two 2 TB drives and six 1.5 TB drives) and it has been working well too. Firewire is noticeably faster than USB (obviously) and iSCSI is even better. So far so good. I've had the first one since shortly after it was released and the Firewire version since shortly after it was released.

I have the Firewire connected directly. Ditto for the iSCSI. Backup with SuperDuper! I have the USB connected to my older Airport Extreme for Time Machine backups. I just switched the configuration that way and so far it seems okay (~48 hours).

I also have a NetGear ReadyNAS Pro (6-Bay). I like the look of the machine and the fact that it connects via Ethernet.

However, there is a BIG problem with the ReadyNAS Pro (and NV) if anyone wants to use it via iSCSI or TimeMachine - it is limited to 2 TB backups. Yes, you read that right, if you want to backup something more than 2TB with TimeMachine or iSCSI (perhaps others, since they say "etc" after listing them) you CAN NOT backup more than 2TB. Given our Mac Pro has four 1.5TB drives RAIDed with 3.5 TB of data, the ReadyNAS has been a big disappointment (and even more so in that it took them 4 months to figure out it wasn't something with the OS or the backup software which meant it was too late to return it).

In April (2009) they said "RAIDiator 4.2.6 road map is to support 2TB+ volumes for iSCSI/TimeMachine/etc, I don't have a timeframe for that release yet."

Needless to say it is the end of August (2009) and still nothing. I bought this end of 2008/early 2009 and so I've had this really nice looking, expensive but useless device for 8+ months.

So, until NetGear fixes these problems I would urge MacInTouch readers to stay away from the ReadyNAS devices because although the advertise 6 bays holding 2 TB drives each for 12 TB of storage, you really can only back up 2 TB via iSCSI or TimeMachine, "etc." (rsync may be a solution, but I prefer *simple* and built-in so that you can have more confidence in it still working after upgrades, non-terminal users etc).

You can buy a several external 2 TB drvies for a lot less and use them in the interim until/if NetGear fixes this.

(See here for more: http://www.readynas.com/forum/viewtopic.php?f=71&t=27531)

Aug. 21, 2009

item.98009

Gregory Weston

Larry Nelson says:

"I bought a Guardian Maximus RAID-1 system from Other World, to use as a backup for my music library. It came with a backup program called Data Backup, so I used that to copy the music library from its home on a LaCie 500GB USB-2 drive to the Guardian. This took about 14 hours for my 460GB library.

I wanted the RAID copy to be the working library, so I went through the steps to have Itunes use that. That proceeded normally -- I've done this before -- but when I tried to play a song, Itunes couldn't find it. If I navigated to the track manually it would play, but I don't really feel like doing that for 23,000 tracks."

The correct way to move an iTunes library - presuming it's okay at the end for all the media files to be in a single folder tree - is to use iTunes. You tell it in the preferences where you want your library to be located, and then you consolidate the library. And wait.

Moving an iTunes media collection to another volume via by any other means and then getting iTunes to recognize the files in their new location is usually not worth the trouble. It may be possible that it's as simple as copying the library xml file, editing all the media file URLS in that copy via global search/replace to point to the new home(s), creating a new/empty library file in iTunes and then importing the XML. But in practice it seems there's always some need for manual tweaking. I'm not sure what "the steps" were that Larry followed, but if it wasn't something along those lines, they weren't the right steps.

item.98025

Chris Minnella

After readings others' experiences with Drobo, I thought I'd add mine. I purchased the FireWire (gen2) Drobo a few months ago. I'm using it attached to a AirPort Extreme base station [via Ethernet?] as my Time Machine volume for two laptops on my home network. I followed some instructions I found online to create the sparse volumes necessary to allow Time Machine to work correctly with Drobo (fixed expansion size). When I first got it working I had a little trouble figuring out what password was required to mount the disk (AirPort Extreme password), but since then I haven't changed the configuration. Early on, there were two occasions where the Drobo inexplicably went offline. I had to reset the Drobo and the AirPort Extreme to get things back up. Since then (knock on wood) the whole setup has been working perfectly and I've been very satisfied.

I acknowledge the closed nature of the Drobo storage format, but aside from simple mirroring, I think you'd have a hard time extracting useful data from any RAID setup without the correct functioning controller. Also, for me, Drobo is just one tier of backup. I also perform Superduper backups to a different set of hard drives connected to my old G4 in the basement.

item.98062

Greg Chesney

In addition to using SOFTRAID on my (Velociraptor 300GB) MacPro system disk, I built a generic PC with an Areca RAID card with 4, 1TB WD-green drives in a RAID5 configuration. This box runs Ubuntu and is the file server/backup for my home office.

A 1.5TB bare drive sits in a Thermaltake eSATA (toaster I call it) and this does Time Machine. All works very nicely.

I also use iDrive (recommended on MacinTouch) after being sorely disappointed with both Mozy and Carbonite.

Overkill? Maybe, but I sleep very well at night.

Aug. 22, 2009

item.98100

Skot Nelson

Re:

I acknowledge the closed nature of the Drobo storage format, but aside from simple mirroring, I think you'd have a hard time extracting useful data from any RAID setup without the correct functioning controller.

This is essentially true, and is yet another reason for that age old adage: RAID is not backup, in and of itself.

Generally speaking you'll need the same hardware/software controller to recover your data for anything that's striped.

I do prefer RAID devices that don't introduce proprietary formats for mirroring. If the RAID hardware fails, you can always just pull a drive and install it direct.

item.98143

Ray Thompson

In response to comments about Drobo.... I had a USB Drobo running for about 14 months, then switched to 2nd generation FireWire about 4 months ago. I've had no problems whatsoever, and have just purchased a 2nd Drobo to back up and store in my safe deposit box! I'm using these strictly as archive drives and I have stuff saved as far back as 1977 from my first Apple II! Where do you see names of individual drives to worry about? I see just one drive on the desktop! As for the Drobo utility, never had to use it or worry about it! A friend has worked in parallel with me; we bought our initial drives exactly the same and he has had two of his drives fail, but Drobo handled it! Go figure!

Dec. 30, 2009

item.106900

Gregory Weston

Randall Voth comments:

Why, oh why are "normal" folks using RAID at all? RAID is only useful at an enterprise level to keep the system on-line while replacing a failed drive.

I get the feeling that people are spending a year ripping all their CDs and DVDs onto a RAID device in the mistaken belief that it is some sort of backup. It is not!

If you have 2 x 1TB drives and you set them up as RAID 1, you have a redundant storage of all your data on the same device. If your house burns down, or the device fails, or you corrupt your data or accidentally delete your data, you are hooped. In other words, you need a backup, anyway.

It *is* a sort of backup. Contrary to your comment above, a RAID setup involving actual redundancy will protect you from device failure. It's not a complete backup strategy in itself, but it protects you from realistic catastrophic loss between your scheduled "real" backups.

item.106910

Joe Gurman

This is getting a bit off-topic now, more about RAID vs. other kinds of storage for backup purposes than WD drives in particular, but I have to disagree when Randall Voth says, "RAID is only useful at an enterprise level to keep the system on-line while replacing a failed drive." There are definitely applications for RAID for home businesses and small businesses, particularly video editing shops, and not all RAID is RAID-0 or RAID-1. Some RAID-5 file systems offer snapshots that, like Time Machine, can retrieve earlier versions of files.

That said, Randall's suggestion to store a backup volume offsite is an excellent one. Better is swapping the on- and off-site backups regularly, to make certain the off-site one is not too stale.

The best reasons for using RAID-5 (typically available in any NAS box with four or more drives these days) are: data integrity (because the striping + parity means only a two-drive failure leaves you "hooped"), uptime (because of hot spares and/or failed drive replacement while remaining up), and read/write speed (through striping).

item.106914

Wire

To Randall and those who need it spelled out for them:

Your assumption is wrong and your advice is still correct. RAID1 beats a single drive failure, which is a very likely case of data loss for most users. It is a backup, but a poor one for a lot of less common failures, theft, fire, comets.

I have wondered for a moment why PC makers haven't designed the system to be inherently tolerant of disk failure? The answer must be that the rest of the system ages so rapidly and is so unreliable that even though disks are failure prone, enough other sorts of problems compete for the user's attention and are just about as likely, not the least of which is the constant tech churn which renders any storage format antiquated within 10 years.

Why do people take thousands of digi photos and never look at them? Where do old blog posts go when they die? etc.

Back to RAID, why do people trust the reliability of a dedicated RAID enclosure, esp, Drobo style stuff -- but any RAID5 style gear: What do you do when the container fails?! Power supplies give out all the time. Should you have a backup enclosure? How long will it last? Will the drive be readable outside of the enclosure?

It makes the most sense to me to use individual drives in cheap enclosures mirrored by SW RAID1, or as separate volumes kept in sync by a backup program. This handles the far and away most likely (for me) case of drive failure or wrong deletion. HW RAID1 may be better at disk failure recovery, assuming the controller doesn't impose any special formatting on the drive, but doesn't help me with the all too common erroneous Move-to-Trash-and-Empty-Recycle-Bin operation.

Now if you house burns, or a tree smashes completely through it (happened to me but the tree missed the computer) RAID1 doesn't do a lick of good. But these are quite unlikely events compared to wear and tech churn.

And out of hundreds and hundreds of PCs I've repaired with every sort of problem, just once I've told a user their data was lost forever, and they were annoyed, but it didn't ruin them.

So as bad as disk failures are, and relatively very common, they are still so unlikely that most people have other problems they feel are more of a worry.

What about those old boxes of photos and slides, old tax returns, stocks, land deeds, old cars, 45RPM records, Betamax, blah, blah. The children! Who is going to think about the children?!

item.106915

Skot Nelson

Re:

RAID is only useful at an enterprise level to keep the system on-line while replacing a failed drive.

Your comment that RAID is not a backup is valid, and I've commented on it before.

Your comment above is not true though: RAID provides protection against data loss due to a hardware crash. Hard drives can crash as easily at home as in an "enterprise" level.

Not *all* RAID devices offer hot swap. I personally would be reluctant to use one that didn't, but for home use they may be fine.

item.106866

Randall Voth

Why, oh why are "normal" folks using RAID at all? RAID is only useful at an enterprise level to keep the system on-line while replacing a failed drive.

I get the feeling that people are spending a year ripping all their CDs and DVDs onto a RAID device in the mistaken belief that it is some sort of backup. It is not!

If you have 2 x 1TB drives and you set them up as RAID 1, you have a redundant storage of all your data on the same device. If your house burns down, or the device fails, or you corrupt your data or accidentally delete your data, you are hooped. In other words, you need a backup, anyway.

So why not use the 2 drives this way: Copy all your data onto one drive. Backup all your data onto another drive. Store the second drive someplace safe, offsite. This is not such a big deal because 1TB of music and movies is not going to change much.

I use several D-link 323s this way. They both have 2x1TB drives from different manufacturers accessed as 2 network drives of 1TB each (NO RAID! NO JBOD!). I store one at my dad's place and, as a bonus, he gets to listen to the music and watch the movies. Every few months, I copy the changes to his D-link.

Dec. 31, 2009

item.106951

Travis Butler

While the people pointing out uses for RAID in a home setting are correct in a narrow sense, I think they're missing the broader picture - user psychology. Way, *way* too often, I've heard less-technical users talk about RAID (and almost always in the context of mirroring) as a panacea - as long as they've got RAID, they're somehow protected from all disk problems. And as others have pointed out, this is far from the truth - there are lots of disk issues that RAID doesn't protect against.

RAID can be a useful component of a broader backup strategy, but it is not a sole backup solution. The problem is that most non-technical users think mainly if not exclusively in terms of sole backup solutions; they don't want to think about multiple backups, rotating media, etc. They just want something that plugs in and protects them. And RAID is not that solution.

Time Machine is probably about as close as we're going to get to that kind of solution with current tech. TM doesn't solve the offsite problem; but I think the only thing that'll solve that, in a way that's *actually used* by the average home user, is home broadband improving to the point where you can practically do network backups - without taking days or weeks to transfer 100 GB of data.

item.106965

Robert Whalen

I would agree with the idea of using RAID and off-site hardware for backups, which is what we do at my company. In the past we relied on tape backup systems at each workstation. It was effective and expensive but slow. We also once had the sad experience of having a bad backup tape. That was especially disappointing since the tape backups had given us a false sense of security.

Hard disks are cheap now, we use RAID 1 arrays in each Mac in combination with an external, normally-off disk that is a clone of the RAID 1 volume that we update once a week. For truly critical files (like our financial records) we have off-site disks that are updated monthly. It's not a perfect system, but it limits what we might lose to something manageable.

item.106972

Tony Schaps

All this talk of RAID 5 reminds me of some recent research I did on the topic which found there is a vocal group (with some pretty well-supported arguments) that RAID 3, 4, or 5 are relics of an earlier time with more expensive drives. This group calls itself BAARF, ("Battle Agaist Any Raid F", the "F" standing for three, four or five). Check out their opinions at http://miracleas.com/BAARF/

Add to that another interesting article over two years old: Why RAID 5 stops working in 2009, which argues that we have larger & larger hard drives, but the unrecoverable read error (URE) rate stays static at about 10^14 bits read. Thus, if you have one drive failure on a RAID 5 array with 6 GB of data, during the vulnerable re-build process there's a 50% chance of one URE, and it just takes one to tank the whole re-build, and all data is lost (pending external backup retrieval). Very interesting stuff.

The solution, at least in my mind, is RAID 10 (a combination of RAID 0 and RAID 1) which is not as fast as RAID 5 and uses more disks, but is more fault tolerant. RAID 10 is explained at http://www.cyberciti.biz/tips/raid5-vs-raid-10-safety-performance.html

But I also agree that RAID is not backup. RAID keeps you going in the event of a disk failure instead of crashing down the whole system. A current offsite backup is always recommended, and an occasional tape archive of the offsite backup may also save your butt someday.

item.106981

Randall Voth

Thanks to those who corrected my statements about RAID. My concern is that most "normal folks" may have a large amount of static data and falsely believe RAID is a backup. By "enterprise", I meant business where data must be online at all times, not the home. My business uses maybe 100 gigs of storage and I have several backup methods (swapped offsite Time Machine disks and a nightly clone), none of which involve RAID. Much of the data has been intact since 1988 and has never been lost through countless iterations of Macs, etc. Even the icons are still in the same place as they were in the Mac OS 6 Finder!

But who needs terabytes of info in the home? Those who rip their large CD and DVD collections. My project has taken a year, but it is quite static. I only have so much money to put into drives, so individual drives plus an offsite backup is best. Using RAID would have cost me twice as much for the same storage and I feel that, given only one choice, an offsite backup is more important than RAID.

But when 2 terabyte drives come down in price I will likely take everyone's advice and use RAID 1 in my own device to protect me from a hard drive crash.

item.106984

David Knuth

I would not trust a RAID as the end-all be-all of backup solutions. The only way to ensure a full and complete durable backup is to utilize some kind of storage media that is not prone to device age and failure, as well as to have a dedicated alternate storage setup entirely.

One could go all-out and integrate some flavor of RAID 1+0 into their setup with hot-swappable drives, two banks of n drives, each bank a striped RAID, and one bank a mirror of the other. In terms of active preservation against hardware failure, this option is fairly bullet-proof by comparison.

RAID 5 is also fairly redundant and works well, but as it essentially rebuilds data by interpolation based on other drives in the stripe rather than having a full mirror, I'd be less inclined to trust it over having some form of mirror or striped mirror configuration.

External RAID devices, as another reader pointed out, aren't something I'd trust unless they offer some flavor of hot-swapping and a control mechanism to allow rebuilding the raid. I've seen LaCie "Big Disk" drives with two internal drives go tits up because the hardware RAID controller dies, and without another exact model enclosure, data recovery is very difficult. Also, forget about your data if one of the drives dies. The "Best" solutions I've found are the external RAID boxes that allow hot-swap.

I'd also be rather leery of utilizing a software RAID over a Hardware RAID for performance and recoverability reasons, but that's me.

item.106997

C Z

R. Voth asked, "Why, oh why are "normal" folks using RAID at all?" ... to which I reply: because it may not constitute backup in and of itself but can be part of a valid backup strategy for "normal" folk.

Take a look at the hardware RAID-1, bare drive offerings from Wiebetech or IcyDock.

I use this strategy: two RAID-1 enclosures. One contains a pair of 750 GB drives, the other a pair of 500 GB drives. I use SuperDuper to back up to the 500 GB target in the wee hours of each morning. Time Machine uses the 750 GB target.

Every now and then, I simply pull the second drive from each enclosure and take 'em off-site. On replacement (with an earlier set), each automatically rebuilds the mirror. My machine is completely oblivious to these swaps and no special procedure is necessary for removal, replacement or rebuild. I can have as many complete backups as suit my willingness to afford sets of drives. I have bootable backup as well as continuous archiving. I have backup that is as pertinent to mishap, error and malice, as it is to theft, fire and flood.

And a couple of $30 drive docks make the drives accessible practically anywhere.

Now I'm thinking of adding CrashPlan to the mix for added resilience and up-to-the-minute data availability.

Jan. 4, 2010

item.107046

MacInTouch Reader

I read with interest the article concerning unrecoverable read errors. There seems to be a major flaw in his thinking: if RAID is susceptible to URE's, then so are single drives, only more so, because they have no parity and therefore can't recover from even a single URE. RAID 5 should be able to recover from one, since the data is redundant, and RAID 6 can recover from two, which is why I use only RAID 6. If you want to store multi-terrabytes of data (and I am speaking now of data STORAGE, NOT backup), you really have no choice but some form of RAID. I started out with Apple's old XRAID, which unfortunately is now way out of date (because it uses PATA drives, which are hard to obtain in large capacities, and doesn't do more esoteric configurations like RAID 6), and then I found that Apple's newer solutions are just too expensive.

The original definition of RAID was "redundant array of INEXPENSIVE drives"; unfortunately that thinking has now shifted to "independent (and expensive "enterprise class") drives. I'm skeptical of manufacturer's claims that "enterprise class" drives will really last longer than desktop drives, but they do offer longer warranties. The whole point is that, with an array, if a drive fails, you pull it, replace it, and hey presto, no down time. What other choice is there, if you need to keep 8 TB of data on-line at all times?

item.107067

Greg Chesney

Re Articles cited by Tony:

Thanks. Excellent articles. I'll be switching over to RAID10 this weekend (1) because I can and, (2) because I've done more disk RMAs in the last year than my previous lifetime. For example, I did have a single-drive failure in a RAID5 (4xWD 1TB Green, yes, green), but it rebuilt just fine.

Of course I monitor all my drives with various smartmon-based freeware at home and at work, and I don't hesitate to RMA a drive when it *starts* to show failures, not *after* it's failed. The manufacturers are so used to it that they don't question the reason for the RMA. I have found that the RMA process that each manufacturer has it almost as important as its product. Another reason that I'm sticking with WD for my drives for now.

Happy New Year all. MacInTouch has been my premier Mac rag over the last 7 or so years and hardly a day goes by that I don't read it. Thanks Ric!

item.107073

Dean Suhr

I run a small business server out of my house with several users. I don't need the performance gains offered by RAID. My disk/backup strategy is a hybrid.

I use a 4-disk Drobo as my primary storage for Time Machine and other single file storage (Entourage databases are backed up using Synchronize! Pro X from Qdea early each day). The Drobo avoids single disk failures and is exceptionally easy to upgrade the disk sizes as I fill up the array.

I have a separate single external USB drive that mirrors the Drobo each night. This protects against a Drobo controller or multiple drive failure. I may make this drive the primary backup with the Drobo as the secondary - it seems that the Drobo can be fairly slow at times while it is doing its housekeeping.

And I am also implementing CrashPlan Plus to backup to a remote computer in another county. These secure/encrypted backups are my off-site strategy against power, fire, burglary, etc. I do not pay for their services, instead am swapping remote storage for free with a friend.

Happy New Year all.

Jan. 5, 2010

item.107110

MacInTouch Reader

Please explain the details of the comment that "Of course I monitor all my drives with various smartmon-based freeware at home and at work" - what specific software; how do you monitor?

I have a 4-drive software RAID 10 created with Apple's Disk Utility in a Mac Pro and wonder how will I know if a drive has failed (or is starting to fail) and needs to be replaced? Will Mac OS X warn me before multiple drive failures make the data unreadable? Without adequate notification of disk errors, I have the nagging suspicion that RAID could become a technique to hide disk errors until it is too late to recover your data.

item.107112

Arthur van der Harg

With all the discussion about RAID-5 suffering from UREs on a single disk: this should not be so much of a problem in a RAID array as on a single disk. For a RAID disk a single URE is not a disk failure. An URE is a problem reading a sector, but what's in that sector can be reconstructed from the parity data. There is really no need to rebuild the entire partition. In principle the drive controller could just remap the sector and work with the RAID controller to reconstitute only the single failed sector. Note however that the drive controller needs to know the drive is part of a RAID. I don't know if this is another reason why there's RAID qualified drives: they might have on-disk firmware to handle this situation.

So a single URE might not exactly mean disk failure, but the real problem is when a disk goes bad, like physically failing with a blown controller or running out of remapping space. It could be just a random failure. In my experience though, something *caused* that drive to fail. Could be running at an elevated temperature for too long, a power spike, a sysadmin tripping over a loose tile during maintenance and hitting his knee on the RAID enclosure... Could be just regular wear for older drives. The something that caused that drive to fail likely affected the other drives in the array as well. When a drive in a RAID enclosure physically fails: watch the others closely and make sure you have a) a current backup, b) a spare disk or canister and c) plans to replace either the entire array or all disks in it if other disks show signs of failure. I've been in a situation where two drives failed shortly after one another. It's no fun.

item.107137

Steven Wicinski

I guess I must be one of those weird people who use RAID. I have my PPC G5 set up with two drives in a RAID 0. (Yes, RAID 0!). Why? Because it was cheaper to get two drives than one large drive, and mainly because my large media library wouldn't fit on one 500GB disk, and I really didn't feel like splitting up various parts to various hard drives and trying to manage it all myself.

It all was being backed up to a 750GB external (oh for the ability to put in more than two drives in a computer!) for Time Machine (ignoring the media directory), and cloning the whole thing to a TB drive I bought later (a 'green' WD). I changed this late last year to a Thermaltake 2-drive case through a fast eSATA port (wow, who knew one would have use for those PCI slots or the ExpressCard slot in my MBP). The drives in the enclosure are also RAID 0, because, again, Time Machining up 1TB of space to a single drive just doesn't offer much room for the actual older copies of files, the only reason to use Time Machine.

But, like I said, I guess I'm one of the crazy ones. (And, yes, I know if I lose a drive I lose everything, which happened a year ago, but isn't that why one has backups to begin with?)

Jan. 6, 2010

item.107184

Greg Chesney

To the reader asking about smartmon tools:

smartmon is an open-source project which has spawned freeware (SMARTReporter) and shareware (SMART Utility) for the Mac and free/share (Active Disk Monitor (Windows) which all allow you to monitor SMART params for individual drives, scan them for bad blocks, etc. The features depend on the tool you use.

I use a variety depending on whether I'm at work (forced XP) or home (Mac/VMWare and Linux) and try to stay on top of things by monitoring the output of the tools. Often, if you wait until you have a smart error on boot, it may be too late.

I also use SoftRAID on my Mac and, as Ric has mentioned previously, it provides error detection superior to what OS X does. I'm beta testing the next version and it is fabulous. I have no affiliations with any of the products I've mentioned - just a happy smart use with no downtime.

item.107207

Leif Carlsson

Depending on what data you have and how much of it is changing between backups should be a factor when determinating your storage solution. Also how much time would have to be spent to fix/restore crucial lost data (if even possible) comes into play. And do you have that time available - maybe there is a deadline?

If others depend on this storage, downtime can be costly if several people can't work. So "saving" money on a too cheap solution can be a real bad thing.

item.107230

Andrea Marchini

As I understand it, RAID is aimed at ensuring continuity in the availability of data rather than aimed at security of the same, although it sometimes ay result in a false sense of security for the user.

And the low cost of nowadays drives eliminates the benefit of sparing one disk or two in the array.

Unless additional speed is required, as provided by RAID 0 systems, I can see no reason for adopting RAID. And this is instance as well is going to disappear as SSD drives become popular.

In my case, a RAID 5 enclosure of 4 disks with one spare disk went bad just because, after replacement of the failed disk and the rebuilding of its content by the system, with no apparent mistake, the whole array became unreadable for my Mac. I am trying to recover the (luckily small) part of the data not being backed up.

Jan. 7, 2010

item.107243

MacInTouch Reader

Thanks for the additional explanation about smartmon tools. (Actually, I already use SMARTreporter and didn't realize that this was an example of what you were describing.) However, I thought that there was a large Google study that found that SMART has no predictive value about drive failures.

So again, how will I know if the 4 disk RAID 10 array in my Mac Pro (created using Disk Utility) needs to have a disk replaced? Waiting until there is an evident disk error seems problematic as the array would likely be unrecoverable by the time that a second or third drive failed leading to a software evident read or write failure. Unless the OS or some monitoring software tells you there is a problem while the RAID software or drive firmware does successful recovery, it seems like RAID would hide the drive failures until the data was permanently lost. Am I missing something?

Jan. 8, 2010

item.107351

MacInTouch Reader

MacInTouch Reader

However, I thought that there was a large Google study that found that SMART has no predictive value about drive failures.

From memory (ie, I can't be bothered to look it up right now!), what it found was that SMART does have predictive value, but there are a significant number of failures that aren't (and can't be) flagged in advance by SMART.

So if SMART reckons a drive is on the way out, it probably is; but a clean bill of health from SMART doesn't necessarily mean the drive won't fail in the near future.

Jan. 9, 2010

item.107368

Robert Mohns

A MacInTouch Reader writes:

"However, I thought that there was a large Google study that found that SMART has no predictive value about drive failures."

This misinformation comes up again and again. I'll debunk it again:

Google found that SMART could not predict all drive failures. However, they did find that when SMART starts predicting failure, a failure usually is imminent, especially if the particular error is a scan error, or an increase in reallocation, offline reallocation or probational counts. Here's a direct quote from the abstract:

"Our analysis identifies several parameters from the drive's self monitoring facility (SMART) that correlate highly with failures. Despite this high correlation, we conclude that models based on SMART parameters alone are unlikely to be useful for predicting individual drive failures."

In other words, it confirms what we always knew about SMART. It can tell you a drive is dying. It cannot tell you a drive is truly healthy.

Think of it as like taking your temperature. A thermometer can tell you that you have a fever. It can't tell you that you have high blood pressure, cancer or a broken arm. Does this mean you shouldn't pay attention to your body temperature? Of course not. Merely that cannot create not a full picture of your health.

Here's a direct download link to the paper:
  labs.google.com/papers/disk_failures.pdf (PDF, ~250kb).

For details on SMART factors, see section 3.5.6 "Predictive Power of SMART Parameters", and the Conclusions.

Jan. 11, 2010

item.107465

Mark Thomas

Regarding Robert Mohns's comment on SMART, let me just say that I've experienced many hard disk failures in recent years and not a single one was predicted by SMART. In each case, the SMART status remained "verified" even after the crash. This makes SMART worse than useless - it's unpredictable and misleading; in short, a dangerous lie.

[While hard drives may fail sometimes without advance SMART warnings, one certainly would want to pay attention to those hard drives that do exhibit warning messages prior to failure. SMART Utility was very helpful in this regard for a slowly-failing MacBook drive we had here. -Ric Ford]

item.107473

MacInTouch Reader

[Re: Mohns note]

Yes, replace drives that SMART says are failing, but you can't rely on SMART warnings to avert data loss because drives fail without warning half the time.

Jan. 12, 2010

item.107536

Colleen Thompson

Re the SMART discussion:

Most modern hard drives come with self-monitoring circuitry (the SM part of SMART). There are many attributes that a drive monitors, which can vary by manufacturer.

An important distinction to keep in mind is that the SMART verification in things like Disk Utility (i.e. the only possibilities are "verified" and "failed") looks at all the SMART attributes of a drive, and makes an overall conclusion based on what someone has decided is OK and Not OK in terms of individual SMART characteristics. In my experience, the SMART verification in DU (and TechTool Pro and Disk Warrior and Onyx, etc.) is pretty worthless, except that if it says Failed, then you know your drive is toast. And if a drive is unusable due to bad blocks (sectors), it won't even acknowledge that, usually reporting the drive as Verified.

This is why I like SMART Utility so much. As the developer explains on his web site, http://volitans-software.com/smart_utility.php, this was an unsatisfactory state of affairs, and he wrote SU to look at the SMART attributes and alert you if any errors below the official "failed" threshold indicate a failure in the making. Not only that, it provides an easy GUI for inspecting a drive's individual attributes directly (using smartmontools.)

I have had SU predict quite a few drive failures, and confirm many drives with failing blocks (which would otherwise necessitate a lengthy scan with TTP, further hastening the ailing drive's demise.) On only two occasions have drives failed with no hint of failure from SU; in those cases, I wonder if the SMART circuitry was at fault.

(I have no relationship with Volitans Software other than as a satisfied customer, yadda yadda.)

The only thing that I lack now is a toaster-type enclosure that would include the wiring necessary to pass SMART data, because almost all external enclosures lack that. Anybody know of one? Anybody feel like making one? There'd be a pretty good market amongst IT people and consultants, at least!

[I tried Smart Utility at Colleen's suggestion and found it very informative and helpful. (Some interesting, extra data are provided below the top level of data, which is already much more detailed than Disk Utility, for instance.) Unfortunately, I also experience the same lack of SMART data access with external eSATA drive docks - both an OWC Voyager Q and a BlacX model - which are connected to a PCIe eSATA card in a G5 duallie. -Ric Ford]

Jan. 30, 2010

item.108673

Colin Johnson

I want to tell you about some serious problems with the HighPpoint RocketRAID 4322 card on Snow Leopard.

I bought the card from NewEgg because the HighPoint site claimed Snow Leopard support [but] they know there are issues and yet they continue to claim support.

Check the Apple support forums and you'll see lots of evidence.

In my case there are two problems:

1) Wake from sleep.
If the energy settings are enabled on the Mac, the computer goes to sleep, as does the array.
When the computer wakes from sleep, the array does not.
The only way to initialize a large array (I had 8x2TB) is to turn all energy settings off.

2) Driver lockup issues transferring large amounts of data.
The driver can panic and crash the Mac with large data transfers.

I bought my card in November and have not been able to use it since purchase.
They RMA'd the original card blaming the hardware and of course that did nothing for me except close the avenue to return the card to NewEgg.

Now I'm stuck with a $500 piece of junk :(

Feb. 1, 2010

item.108732

Steven James May

Colin Johnson writes:

"...Now I'm stuck with a $500 piece of junk :( "

NewEgg seems to take their reputation, therefore their customer service seriously. They have not yet let me down. If they advertised the card as Snow Leopard compatible, and it is obviously not, then business ethics will compel them to fix your deal. Have you called them? I would not be stuck with a card that does not work... I would be surprised if NewEgg let the deal end like this.

Feb. 6, 2010

item.109101

Ned Simpson

I have a RAID system that, under Mac OS X 10.5.8, the RAID would be recognized by the system if brought up after the system was running.

Under MacOS 10.6.x, it will not. It will only mount on a cold start.

Both Highpoint and Sans Digital are less than forthcoming in response to my queries.

Have done the standard troubleshooting, permissions, safe boot routines, etc.

Any suggestions?

Model Name: Mac Pro
Model Identifier: MacPro1,1
Processor Name: Dual-Core Intel Xeon
Processor Speed: 3 GHz

Vendor: HighPoint Technologies, Inc.
Product: RocketRAID 231x/230x Controller
HighPoint Web RAID Management 1.6.7 software
Latest drivers installed

Sans Digital TRM5-BSATA to eSATA JBOD/RAID enclosure w/
5 - 1 TB drives in a RAID 5 configuration

Mar. 1, 2010

item.110342

MacInTouch Reader

I have been looking at the CalDigit HDone and HDelement and wonder if anyone has any experience with these. They might be a bit pricey, but I like the apparent simplicity of not having to buy pieces from different vendors to make it work. I am looking for 4-8 TB of high speed storage for day to day work. My Drobo Pro is fine for Time Machine but isn't really fast enough for my work, nor is Apple's software RAID.

So any CalDigit experiences or other single vendor solutions I should be looking at?

Mar. 2, 2010

item.110372

Samuel Herschbein

Re: Ned Simpson's question:

I have a RAID system that, under Mac OS X 10.5.8, the RAID would be recognized by the system if brought up after the system was running. Under MacOS 10.6.x, it will not. It will only mount on a cold start.

FWIW: with a totally different setup, I've experienced eSATA drives not mounting. What works for me is to unplug & plug in the eSATA again.

item.110375

Samuel Herschbein

I'm using an OWC Mercury Elite-AL Pro Qx2 with a 4-1TB enterprise drive config. It's attached to a Mac mini with OS X Server via FireWire 800. As with everything OWC I've bought: very well made, robust, and no issues setting up or using.

I tested this RAID on my MacBook Pro 17" and compared FW800 to eSATA: the eSATA was MUCH faster overall, some writes were almost 2x. If the mini wasn't under warranty, we weren't so poor, and we needed the speed, I'd disconnect the 2nd internal hard drive and pull its SATA [connection] out of the case for the RAID.

May. 27, 2010

item.115102

Stephen Clark

Home's Gen 1 Mac Pro wouldn't boot yesterday; looks like the 2-drive Disk Utility (v11.1) mirrored RAID pair became degraded.
(10.5.8 on two identical Seagate ST3750330AS 750 Gb SATA drives.)

Trying to get the intact RAID member to serve as a Restore source, I repartitioned the degraded member drive. Now the drive I wanted to Restore from won't mount nor is it visible in DU.

Is there some Terminal or other magic I can use to get it to a mountable condition?

None of the drive utilities I have will see that particular drive (Disk Warrior, Data Rescue II, Tech Tool Pro, SoftRAID) nor will it show a drive description in System Profiler. If I move it from one bay to another, the generic description follows along, but is identical to that of the two Intel processors.

Do I have any recourse to get the intact but now solo RAID volume working so I can "clone" that volume to another disk? Or even get that particular drive back to a format-able state, even if the data can't be recovered?

May. 28, 2010

item.115146

Colleen Thompson

Stephen Clark: what happens if you put it in an external enclosure, and perhaps try it on another computer?

item.115164

Mark James [SoftRAID]

Stephen Clark;

Contact support@softraid. I can get you up and running, either with SoftRAID, or the command line. We recover a fair number of Apple RAID systems that blow up, this is not so unusual. Your problem is probably one that does not require engineering to fix.

Oct. 2, 2010

item.122086

Colleen Thompson

Anybody have any experience with Other World's Mercury Pro Qx2 hot-swappable 4-bay RAID enclosure?   http://eshop.macsales.com/shop/hard-drives/RAID/Desktop/
I'm thinking about filling it with 1.5TB WD Caviar Black drives in a RAID 5 configuration. I know these drives aren't officially for RAID, but I've never had any trouble with regular drives in a RAID enclosure (knock wood).

Though a loyal customer of OWC for many years, I haven't had much luck with Newer Tech's Guardian Maximus enclosures (2 of 4 have failed, both failures involving the enclosure and one drive. I suspect the fan or the power adaper, both of which parts are easy to source too cheaply.) So I wonder what others have experienced with this 4-bay enclosure. Newer Tech is part of OWC, or vice versa.

Our Intel XServe's storage is getting tight (it's got an 80GB with the system, and a 750 and a 1TB for data.) And you can't get internal drives for it any more from Apple.

Next Page...


MacInTouch Amazon link...

Talk to MacInTouch     Support  •  Find/Go