Claiming a drive failure rate of 0.25% per year for these devices is absurd. More like 25% failure per year. Do a Google search and see how many complaints there are of one drive after another failing until the warranty period runs out. I'm honestly surprised WD still sells these things.

Why do you think WD touts how easy it is for customers to service the unit themself?

http://en.wikipedia.org/wiki/Western_Digital_My_Book#Morse_code

"The Morse code message written into the drive case is made up of a selection of the words "personal", "reliable", "innovative", "simple", and "design".[17] The first occurrence of "innovative" on the My Book Pro and My Book World Edition features a misspelling and reads "innovateve"."

Google is your friend.

is there a message in morse code on the back of the casing ???
cos it sure looks like morse to me....

anyone care to decode it ???

.25% would be the chance of both drives failing in say a year, but the chance of the second drive failing before you replace the first defective drive is more like .0025% if say the it took about 3 days.

And yes using RAID0 would double the chance of failure. It also adds other issues too that if say in a few years the enclosure dies and you pull the drives off you need a compatable chipset in the new device to be able to pull the data off.

$649 / 1.5 (U$D to Euro) = 433.33.
Add vat 433.33 * 1.2 = 520.
Add a exchange-ratio insurance (although, these days, it's more likely the Euro will get to 1.6 rather than 1.4....) and you'll get to 550 Euros. Not 699.

But we shouldn't complain. Brits, as usual, get the first prize for exchange-rate ripping.

USA 4 Africa??? Nope, it's euro-commies for USA.....

While I haven't owned one of these, I certainly won't. I was one of the poor saps who bought a MyBook World and had it fail after a month (like the rest of them did). Western Digital replaced the dead hard drive and then that one quit after a few months.

The IT guys at my office bought a few of the 1TB versions of this device and none of them lasted six months.

I figure it's a serious design flaw Western Digital doesn't bother to fix. The two drives are packed into a tiny case with an integrated power supply and just one 30mm fan. My bet is the drives just cook.

The only thing this device would be useful for is backup, but if you can't trust it what's the point?

The interface on these is far too slow; even the MyBook World with a supposed Gigabit Ethernet interface was far slower than a HDD in a $15 USB enclosure. Don't even think about sreaming video off of one of these.

Hopefully they stopped bundling that god-awful Mionet malware with this thing, too.

RAID 5 disks will get very similar but not identical workouts. Show me the evidence that IDENTICAL workloads will cause a significantly higher chance of failure compared to similar workloads.

Disks fail due to a variety of reasons, but even in the same batch you'll never get two absolutely identical drives. There will always be small variations in the mechanical and electrical tolerances of components, the precision of assembly, the faulty sectors that are mapped out, the amount of heating and cooling that the drive undergoes, handling during shipping etc. Trying to prove a statistical difference between failure rates of disks under similar and identical workloads is going to be more than a little tricky as they're never going to be as identical as you seem to think.

There is statistical work to show that you can reduce the chance of a two drive failure by picking drives from different batches or manufacturers, but I've only ever seen it applied to large RAID 5 arrays where the risk of a second drive failure is much higher.

That's RAID-5. The workloads on the drives aren't identical; they do mostly come from the same batches, but they read and write different things, unlike RAID-1.

Too simplistic an analysis of drive failures. RAID 5 arrays are MUCH more at risk of a two drive failure (in, say, a 5 drive array once a single drive has failed then the whole array is at risk if any of the other 4 fail before the degraded array rebuilds onto a replacement drive. There's slightly less than 4 times the probability of one of these 4 drives failing than the second drive of a RAID 1 drive array also failing before replacement of the failed drive. RAID 5 arrays are however traditionally built from drives from the same batch and very rarely is this a problem.

The rates of error may vary between batches, but the time to failure of each individual drive within a batch is highly variable, even with identical workloads.

... is extremely high.

Why? Because both drives are going to be from the same series, using the exact same batch of materials, etc.

On top of that, both drives in a RAID-1 field will be doing the same tasks - reading at the same time (the controller needs to verify that both give the same data) and writing at the same time (because that's the point of a mirror).

If you have two identical drives with identical workload, take a guess about the probability of both failing in a same short period of time.

So, if you need a mirror, either get drives from very different batches, or use drives from different manufacturers.

Because of the need for an MS Windows tool and because of the lack of a 1 Gbit/s ethernet interface, I won't waste my money on this one.

RAID 1 decreases the odds of both drives failing, but RAID 0 *increases* the odds you will lose your data. If there is a 5% chance of 1 drive failing, there is a 10% chance of either drive failing. With RAID 0, if either drive fails, you lose everything.

If you need the storage capacity of RAID 0, use two independent drives. The chance of failure is the same as for RAID 0, but you only lose the data on the one drive instead of both.

Only about $450 in the U.S.

The thing works fine under Linux. I used the windows tool to reset it to 2TB RAID-1, partitioned/formatted with linux, no problems at all.

1) The chance of a single drive failing IN A YEAR is about 5%, the chance of two drives failing in the same period of a week is MUCH less than 0.25%
2) RAID 0 gives you MUCH higher throughput than a single drive, but as you're likely to max out USB or even FireWire 800 with the single then you won't see much advantage (eSATA may prove more useful)
3) FireWire 400 to 800 adaptors are available as well as third party 400 to 800 cables. The whole point of having TWO 800 (or 400) ports is that you can CHAIN them. You can hang up to 127 drives off of the one controller.
4) XP and 2000 are limited by the MBR boot sector descriptor that they use. They simply cannot understand bigger single volumes (although I believe you can make them span multiple 2TB volumes in a software RAID 0 configuration). This is not WDs fault.
5) MacOS defaults to GUID partition tables, which can be much larger than 2TB. Vista and now Windows 7 also understand GUID tables.
6) WD looked at what their buyers were using and concluded that more MacOS users wanted this product than Vista/Win7, so they pre-format it as HFS+ rather than NTFS. Reformatting is pretty trivial.
7) Putting an ethernet controller in the box is far from trivial, and it compromises performance badly. Cheap NAS controllers max out at about 5-6MB/sec, even on Gigabit ethernet. USB can push 30-40MB/sec, Firewire 800 about 75-80MB/sec and eSATA about 150MB/sec. You need a fairly quick CPU and lots of cache RAM to make NAS work well, and that costs money. 4TB @ 6M/sec? Really?
If you REALY want 4TB of network backing store then for little more you can buy 5x1TB drives and a Thecus 5200 Pro, which gives you 4TB of RAID 5 storage available as either NAS, or much faster iSCSI SAN over Gigabit ethernet.