Anybody who is confused about what MTBF means *cough*
ribbon13 *cough* should read about it:
http://www.storagereview.com/guide2000/ref/hdd/perf/qual/specMTBF.html
The MTBF of the Raptor might be 1.2 million hours, but this is often totally theoretical, and like
Matthias99 was saying, it is the representation of many drives running for a short period of time.
For example, say they shipped 30,000 Raptors. One year later (8,760 hours) they had 219 dead ones returned to them.
30,000 * 8,760 = 262,800,000 hours total service for 30,000 Raptors
262,800,000 hours / 219 blown up Raptors = 1,200,000 hours MTBF
That's certainly not bad, but think about it. Even if over 20,000 of those drives saw constant use, that still means you have a solid 1% chance that your drive will die within the first year. It really isn't quite as impressive as 1.2 million hours first seems.
The fact is that no matter how it is calculated, actual reliability is probably far worse. If the MTBF comes from the lab they certainly do less than a full year of stress tests. If the MTBF comes from RMAs they certainly do not get 100% returns on faulty drives.
The chance of failure increases rapidly over time. If a drive isn't DOA or acting funny the first time you power it up, it'll probably last 6-12 months, but who knows how the Raptor 74 will be holding up after 3 years? It's certainly not going to last until it's totally obsolete.
Then again, I do still have two Quantum Atlas 10K IIs (from early 2000) and a Viking II (from early 1998) which are still doing just fine. If WD is building Raptors as solid as Quantum/Seagate builds 10K drives, I suppose a decent number of them will live till they're obsolete.
The specs on the Raptor also say:
Contact Start/Stop Cycles: 20,000 minimum
That's lower than average ... and this is another quasi-legitimate statistic. Obviously there would be a good chance of damaging the drive if you spun it up/down and crashed the heads into the landing zone 20,000 times in a row.