I would like to let you know that the trim command will not work if you use raid, this is a hardware limitation; I might need more information regarding the "RST driver".
Windows 7 support is capable of handling TRIM natively so this is a good step in your configuration if you are planning to use an SSD.
Adding to Mauriceb's reply, Intel has indicated intention to support TRIM for SSD RAID, but it's not clear when that will come. New RST version 10 drivers are promising (available on the Internet, but not officially released yet). Early reports indicate significant performance improvement on SSD RAID arrays that are "seasoned" (very few or no remaining blocks of memory that haven't been written to and erased at least once). This is a sign that RST 10 might support TRIM. Lots of good info on this at a new site dedicated to SSD: The SSD Review
RAID-0 is always a risk due to lack of fault tolorance. The failure of one drive causes failure of the entire array. Since an SSD already has much lower access time and latency compared to rotational disks, the risk of RAID-0 stripping is difficult to justify. If squeezing every bit of performance out the technology is more important than possible loss of the array then you may be a candidate for RAID-0.
I'm using two X25-M G2 80GB SSD's in a RAID-1 array. RAID-1 "mirroring" is all about fault tolerance since you can lose/replace one drive and keep running. Some might say SSD also makes RAID-1 more difficult to justify (in terms of expense) because solid state should inherently be much more reliable than mechanical disks. However, electronics can also fail and SSD is still a relatively new technology and my data is very important.
RST 10 does not support TRIM. The most likely reason people see better performance with RST 10 in a raid configuration is due to the rumored RAID Volume Cache Size Increase to 16 MB of System Memory.
By using NCQ a single Intel SSD can execute up to 32 commands in parallel. If you don’t need added bandwidth raid 0 offers no advantage for non-enterprise use. You face a disadvantage when the drives start to degrade and TRIM can't kick in.
redux, thanks for the additional info. So, boost in cache size and parallel execution is helping to mitigate the negative impact of not having TRIM for RAID SSD. Makes sense. Church bells would be ringing everywhere if RAID TRIM support were discovered in RST 10. LOL
As for RAID-0, I couldn't agree more. I wouldn't be caught dead using stripping in the enterprise either unless it were in a RAID-10 or 50 configuration. I think RAID-0 is a holdover from the rotating disk where it was a little more justifiable for performance reasons if you didn't care about losing data on the array. But I do enjoy reading the experiences and benchmarks of the thrill seekers willing to use RAID-0 anyway.
I want to add, after heavily using a RAID-1 SSD array for almost a year, the effect of lacking TRIM support will probably show up in CrystalDiskMark or other tools, but it's not yet noticeable to me in everyday use. Response time still feels blazingly fast. Might have time to do some actual performance measurements over the holidays...