More on CPU specs of the ReadyNAS Pro

fastfwd wrote:

Single redundancy for me. The "-T" test reads from the portion of your main RAM that Linux uses as a disk buffer; it just tests how fast Linux's disk-cache algorithm running on your CPU can read data from your RAM. The drive isn't even accessed during the timed portion of the test, so nothing about the drive's performance -- including the size of the drive's onboard cache -- should affect the result of the "-T" cached-read test.

Well 4 drives should do a little better than 3 and one of my drives is a cool model (an odd 5900rpm I believe), but it seems hard to believe that is all of the difference. Could you post the output of hdparm -i /dev/sda? I'd like to see if there is anything special your drive can do. Maybe the fact that I have 3 different drive models triggers some poor md behavior due to cache size mismatches or something.

One thing odd I noticed is that when I run the test of /dev/c/c my USB drives all wake up. I've got 4 of them attached to a USB hub. Maybe I should try disconnecting them all?

steve