Bob, have you tried getting the data back? I must say I'm surprised that you've a Maxtor in use -- over time you made a number of "rude" comments about the things!

In all fairness to Maxtor I've not used enough of any brand of disk to get useful reliability data. All manufacturers can and do get it wrong from time to time. Our son bought a number of IBM drives a few years ago very cheaply in SIngapore and they ALL died. Yet IBM kept honouring the warranty, I think in the end he swapped them 3 times and still they kept dying. Must have cost IBM a bundle in air freight between here and SIngapore.

In this instance though I didn't build the system. Given the credentials of the system integrator I just accepted whatever they used and so far I've had one memory stick have a contact problem that they fixed in 30 minutes and then this disk problem. I think the system is still under warranty so no doubt the drive will be replaced at no cost to me. It's just the loss of time that bugs me.

As I said nothing vital has been lost, and I suspect I can pull all the data off it, just finding a spare 400 GBs is the hard part. And then I've got to rebuild the RAID and restore the data, wish those CF DIs weren't so damn big.

Anyway I'm a glutton for punishment, raced out and bought another 3 Maxtors, two SATA and one IDE, at $135 for 300GB they're hard to resist on a cold Monday public holiday.

But I agree with Buster's advice. The Samsung and Hitachi drives do sound like they're more solid. Well I haven't tried a Hitachi in a long time but I have a Samsung in a Shuttle and it's ultra quiet.

Bob.

Personally, I am thinking about getting an Intel SS4000ENA Entry Network Attached Storage System and four Hitachi Deskstar 7K500 500GB 7200 RPM SATA 3.0Gb/s Hard Drives. Both seem to have had good reviews. That way I can run RAID 5 - not sure how risky that is based on general HDD performance.

RAID 5 is good for reliability, you need more than 1 drive to fail to loose the lot.

Xander,
i have been looking into the same set up, does anybody know if it will be fast enough for the CFDI's. theorecticly should be but i have heard some of the NAS actuall thru put isnt what it should be.

Remember that Hitachi bought the IBM SCSI deckstars which were failing faster than they could ship them. They had a 5 year warranty. I had two of them thinking that IBM would be reliable.
JJK

I have had only 1 drive ever fail on me and it was a Maxtor. I have used several Samsung and Seagate drives with no problems. For any device thought I imagine it is only a matter of time. IMO if we get 5 years or more out of any one drive, it has served us well.

Chad

The IBM75GXP DeskStar (IDE) did indeed have serious problems, but it was only one of many hard drives that IBM had at the time. All the rest were quite reliable, but that one bad drive nearly put them out of business. Too bad, because it was IBM that developed all of the cutting-edge technology that eventually ended up in other manufacturer's hard drives. I have an IBM 14GXP drive that has been in daily operation since 1998 and it's still running strong. I have a bunch of Hitachi (IBM) 250 and 400 gig drives that I pound pretty hard every day and they continue to work flawlessly.

I did use Western Digital for a long time, but had a rash of failures that kind of put me off of WD for now. I use Hitachi and Seagate drives and wouldn't touch a Maxtor for any reason..I've had really bad luck with Maxtor in the past, mainly drives that were flaky right out of the box.

John

RAID setups (where there is redundancy) often appear to perform poorly in workstation duties. Many are rigged and resourced to hand-out hundreds of megabyte files per second, but really show trouble if you try to pull off a single ten, hundred or many thousands of megabyte file runs or streams. Such small sized high volume devices are better tuned for enterprise file, messaging and database servers. It doesn't make them junk overall but what us offline editors need to do is to look for workstation class nearline storage.

Enterprise units have a rating of 100's to 10's of 1000's I/O ops per second and you mate the appropriate drives to that interfacing or backplane. Workstation units talk of the number of streams and sustained data transfer rates. OK, you also need to mate with an appropriate drive or set of drives as a resource.

From my investigations, my personal favourite base RAID level is RAID-10. Striping and mirroring. In some implementations (the feature adoption is optional) you can double up on the stream support as the array will split the duties alternately across the mirror and then play catch-up. So if you get 180MByte/sec sustained write and 250MByte/sec sustained read, you can do either concurrently without being held up. The reality is usually less than the theory, but many RAID solutions are responsive without resorting to solid state technology.


My experience is that the modern Seagate ATA/SCSI drive is the most reliable, followed shortly by the WD varieties. I'd then prefer Samsung or Fujitsu if neither of the other two were available, but I would ultimately expect them all to fail, or be capable of failing.

We are not encouraged to keep our money at home, but to instead use a bank. So our valuable information may end up entirely served by storage hosting servers on the Internet. Once we've got enough bandwidth and the right tariff to exercise the use of it. Until then we walk the tightrope of information storage with multiple hard disks (internal, external, portable, network-based, tape-backups - the lot).

I agree though. It is sometimes better to enjoy reliability ahead of cost per gigabyte. At the point when the glass substrate was replaced by plastic for the platter medium, I've a fairly good idea that reliability took a downward trend. The digital world has a habit of driving things to be cheaper. It matters less when your competitor "follows suit".

Despite a problem one time with a Western Digital disk, that is all I have ever used. I still have my first 40 MB Western Digital hard drive in a box somewhere.

I would never take a chance on a Maxtor.

John

If I can be assured though that the reliability of SCSI drives isn't just an urban myth

It is an urban myth. Drive electronics are a distinct and separate entity from the physical mechanisms of the drive itself. The only thing I can think of that might interact with reliability is if there was a big enough cache to reduce the number of head moves, or the "violence" of those moves. For instance, you can download firmware for many drives that makes them operate more quietly by slightly slowing down the movement of the heads. This in turn might make the drive slightly more reliable.

If Storage Review has reliability information based on testing or consistent user surveys, that would be a good way to get a reliable answer to your question.

In my own personal experience, both as a user, manager, and a consultant, hard drives from all manufacturers are amazingly reliable. I have seen posts in various forums slamming each and every manufacturer ("I'll never buy Maxtor again," "Western-Digital stinks," and so on). Don't ever listen to any individual story, no matter how horrible.

CD-ROMs and DVD burners are another story. They are by far the least reliable component in a modern computer and are seemingly guaranteed to fail.

My advice is to buy the cheapest hard drives you can find, and not worry much about the manufacturer.

And finally, if reliability is the question, then RAID is the answer. As you've already found out, it reduces the disk failure to a minor inconvenience rather than a major event.

SCSI drives have always had higher reliability.

Not because of the interface, but because they are built better.

Designed for higher reliability in corporate server applications.

This costs serious money unfortunately.

SCSI drives have always had higher reliability.

Interesting. I stand corrected. I found this abstract of what looks like a pretty scientific study, that supports your statement, and refutes mine:

More Than an Interface---SCSI vs. ATA

>> In my own personal experience, both as a user, manager, and a consultant, hard drives from all manufacturers are amazingly reliable. <<

A point I agree on whole-heartedly when IT managers where I work place so much emphasis on requiring SCSI. We have a situation where we are constantly running out of drive space and an upgrade to our servers is seen as too expensive. We have been asked to move our extensive Outlook folders (100+ employees) to local drives to save space. "Back them up to local DVD" is the response in case you're wondering...

And the suggestion of moving to large capacity CHEAP SATAII as a solution is met with loud guffaws as being too unreliable.

Conversation from me to IT "experts"

Q: "Do you back up frequently?"

A: "Yes"

Q: "So if a SCSI drive does come crashing down it would simple be a case of restoring."

A: "Well, yes. But we run a redundant array mirror and so we can usually keep going without requiring a restore."

Q: "So if you instead used Raid0+1 on large capacity SATAII drives then the same procedure would work?"

A: "Well, I suppose..."

Q: "And given the speed of SATA II RAID 0+1 wouldn't you think that the performance difference is almost immeasurable? Not to mention the savings?"

A: "Well no. There's no real discernible difference for the end user...."

Q: "So why aren't we doing that?. I'd like to put my Outlook on the server instead of local drives because we're out of space."

A: "But SCSI is more reliable........."

Meanwhile, back at the ranch, MOST end users are forced to use Outlook on local drives and don't back up. Reliable? Go figure...

In closing, I wonder how many of us *rely* on ANY drive to look after our veggies. Or perhaps we also diligently and regularly tuck those away with our raws?

Backing up Veg files is simple, backing up media files is not and that's the problem. Sure so long as you have the project files all your work can be rebuilt but a large part of the work on some projects is capturing and rendering intermediate files. In my current project there's 8 hours of source media so recapturing costs a day.

Actually there's quite a trap to be aware of. Just the veg files might not be enough. As others have noted without the capture log files they can be largely useless to you.

Fortunately in my case I have backups of the captured media, phew.

I do agree about your comments about SCSI and storage capacity, not only are the drives expensive they hold way less data so that means in my case I'd need at least 10 drives, my PC has the space and can take a backplane to suit but the cost of 1 TB of SCSI is very scary. I think there might be other avenues that cost a bit less that'd yield the same reliability that I should consider before laying out that kind of money.

Bob.