> IDE channels tend not to have the extra bandwidth you see on a SCSI > controller. One drive can pretty well saturate the channel's > available capacity, resulting in a performance hit. Make one the > primary master and the other the secondary master. my understanding is that IDE drives have a lot less 'smarts' built into the drive itself, than SCSI does. IDE used to be cheaper, because it used simpler controllers and offloaded the control logic to the CPU. as such, there's a lot more 'blocking' operations that happen over an IDE bus -- the control logic issues a command, and has to wait on the results of that command, before issuing the next one. (even if the next command is to a different drive on the same chain). SCSI is smart enough to issue multiple commands to multiple drives on the same chain, and deal with the results in whatever order. as such; SCSI vs. IDE is neck-and-neck when doing a few simple things (like the 95% of end lusers who read mail and www, and maybe play a game that mostly gets loaded into memory anyway); but SCSI wins when doing lots of tasks simutaneously (like on a server with multiple users; or a workstation that's doing a squid cache/dns cache/kernel compile/ftp download). these days it probably doesn't cost much more to make the silicon for a real SCSI controller, than an IDE controller; but the big storage companies have gotten into the habit of charging 2x for SCSI over IDE, and the consumers paying it, and economies of scale contributing to make IDE cheaper. :( life would be a lot simpler if we used SCA drives for everything; but at the moment it's cost-prohibitive (even tho it doesn't need to be). oh well, what would we do if life made sense? :) Carl Soderstrom. -- Network Engineer Real-Time Enterprises www.real-time.com