320gb WD SATA2, AHCI
1.96,1.96,gfs1,1,1293773837,6584M,,709,98,56636,4,26136,2,2395,97,58757,2,192.5,1,16,,,,,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,14246us,224ms,352ms,14567us,15230us,520ms,267us,610us,644us,237us,14us,42us
Intel X25-M, 2nd gen, AHCI
1.96,1.96,gfs1,1,1293774560,6584M,,1026,99,101777,9,75787,8,2108,99,313987,14,15842,123,16,,,,,10180,27,+++++,+++,10635,27,10204,26,+++++,+++,8313,23,7808us,320ms,389ms,4258us,1271us,3544us,7230us,179us,3124us,4890us,42us,24047us
meta-data=/dev/sda8 isize=256 agcount=4, agsize=6318564 blks
= sectsz=512 attr=2
data = bsize=4096 blocks=25274253, imaxpct=25
= sunit=0 swidth=0 blks
naming =version 2 bsize=4096 ascii-ci=0
log =internal bsize=4096 blocks=12340, version=2
= sectsz=512 sunit=0 blks, lazy-count=1
realtime =none extsz=4096 blocks=0, rtextents=0
IO Wait has almost disappeared for the series of machines we replaced. After a number of lengthy calls to Intel, we put these with XFS (journaled filesystem), on some very busy MySQL servers. I'd say we went from 10k RPS to a little over 12k RPS measured after 14 days.
Intel's comment was regarding the degradation of the drive. They claimed that we would easily see a number of years of service before we would have any problem with block rewrites. As the systems are 50% full, the duty cycle was claimed to be 5+ years. We debated Ext2/3/4 versus XFS, but, almost everything we do is XFS already, Intel claimed even with journaling, we shouldn't have a problem.
I don't have before/after benchmarks but I can tell you we built a beefy server with an SSD because our SCSI disk based server was falling on it's face and now the load is very minimal.
Were you on a SCSI RAID before? And are you on a single SSD or RAID config?