Jose wrote:
There are 2.5" SCSI drives. But damn expensive and rare.
And, they claim that you have the advantages of SCSI with lower price. Your're right 70 plus the IDE drive will be cheaper!!
As for the advantage you mentioned SCSI has, sounds interesting are you sure it's a hardware thing in SCS HDs?. If not this combo probably has it, since you have to buy a card like that one for each IDE device, there is no arbitration goin on. .
You caught me- hard to find a good 2.5" SCSI drive now. However, a recent review of the drive industry I was reading noted that everyone *is* interested in 2.5" movements for racks and blades - though those'll probably wait for SAS in the SCSI camp, I guess.
TCQ is a feature found in the (on-drive) controller of basically every SCSI drive made today. So far as I know, only IBM (now Hitachi) have tried to implement it on ATA, and I'd be doubtful if software supports it. (FreeBSD has an option to enable it on ATA devices, with some warnings; I assume Linux is the same, and I'd be surprised if anyone knows what Windows does.) Since it's barely a standard on ATA, I doubt the 'bridges' know how to deal with it.
Yes, performance is sort of comparable with the bridge in place, but it's still some insanely complicated hardware (isn't it all, these days?), and what it basically proves is that, if you spin a drive really fast, you can't help getting good performance even if optimizations like TCQ are disabled. You never know what the controllers on ATA drives are up to these days, either - multimegabyte caches, spiral zoned recording... adding the extra bit of "intelligence" to speak SCSI is nothing vs. the complexity of the drives themselves. This is why I think the world will slide back in that direction once SAS launches; there's no reason *not* to ditch ATA in favor of the relatively simpler, less kludged-upon standard. (Sort of how UNIX has come back against Windows; suddenly the 'obsolete old beast' starts looking like less of a headache vs. last year's 'hip new thing.')
