tweakerz.net did a great counter argument, proving without a doubt that RAID 0 improves speed for many people in many cases
people influenced by storagereview and anandtech would bring up irrational arguments like cases used by tweakerz were not typical of desktop users - I didn't know copying files and multitasking were not in the domain of desktop users??!?!
one of the main problems with desktops is hard drive thrashing in which no useful work is done and it takes 5x,10x,100x longer to finish an I/O job (similar to process switching on the operating system side). With today's single drive systems and physical perfomance limits of hard drives, hard drive thrashing is a very real reality. It is easy to thrash a hard drive - copying a large file to the same hard drive in win98 is enough.
RAID 0 prevents thrashing by giving the computer extra breathing room. MANY non-power, non- CS/CE/EE/Math/Physics/Psych desktop users have two dozen or more processes running, many unnecessary, some malware, an outdated AV, and maybe a virus or two. The OS is constantly context switching, paging, and caching to the hard drive to make good use of memory but with the overhead the system is slowed to a crawl. The user can improve speed by adding more memory or by using RAID. By adding memory, less stuff needs to be driven out of cache. With RAID 0, caching out is faster, as well as several other improvements. Thus RAID 0 not only helps power desktop users, but also grandmas and posers out there.
For people with 256MB of 1066 RAMBUS memory or less from the Dell deal, updating the memory is NOT an option. Therefore the only option is RAID 0.
RELIABILITY
---------------
As far as reliability is concerned, RAID 0 will actually improve the lifespan of the individual RAIDed hard drives verus single drives.
Manufacturers list MTBF, MTTF of a device but have you ever wondered how did they come up with those numbers? It is an inexact science with a lot of exaggeration and no standards. My professor at UCSD, a visiting researcher to IBM storage labs, did not even know.
I fear that some manufacturers test the mean time before failure by letting the hard drives run idle. Measuring the MTBF of an idle device is very misleading. Many things in the real world have their MTBF stated with respect to workload. For example, a relay switch life is measured in how many switches before the realy dies. Because a relay is always on/idle in one state, its idle MTBF can easily be 100,000 hours, or as high as 1,000,000 hours when not switching. But when switching, it's MTBF can be as low as a day.
With RAID 0, the workload on the hard drives is reduced because the I/O jobs are finished faster with less thrashing. Less work in ANY setting and in ANY industry translates to longer life.
CONCLUSION
----------------
In conclusion, RAID 0 is FOR desktop users. The simple Markov models that people reherse so often do not do an adequate job of predicting the MTBF of RAIDed systems. The MTBF should be measured in gigabytes transfered from the beginning of the platter to the edge before failure or bad sector encounterd multiplied by the mean time for each transfered gigabyte in non-server environments.