OK, just an update in case anyone is interested (it doesn't seems so
The RAID 4 and RAID 5 give me exactly the same results. The sequential reads are very fast, reflecting the striping, and the sequential writes are about half of seq, writes, reflecting the fact that the parity needs to be calculated.
The one thing that bothers me, is that the writes should be a *little* faster. I am now in doubt whether the parity is being striped, or is only one disk being used for parity and that;s what's slowing down the writes.
Mcleodnine, you seem to know your stuff when it comes to RAID. What is your opinion on this?