otheos
03-07-2001, 08:26 AM
IDE Raid is not bad (Raid 0) for performace but it really kills multitasking. if IDE has a disadvantage again SCSI in mutlitasking, wait untill you see IDE RAID. It's hoplessly slow!!
IDE RAID 0 is good for A/V editing where sequencial read write matters.
Put the OS on it and you're dead. If you think of it, all the fuss about how multiple drives are beneficiary to performance is cancelled by IDE RAID 0 since you end up with one big disk where all heads are sync. Desperate.
I know Sandra lovers will be excited with the benchmarks but try moving a big file around in the array. I've done tried it:
it takes longer to move/copy a file from one partition of the array to another than to copy/move the same file between the two disks if they are seperate. It makes sence. It's the same as having one disk and two disks. No matter if the one disk is almost 2x faster than the two appart.
Try this. Put the OS on the array and in startup put 10 programs.
do the same on the two disks apart with the OS on the first disk and the programs on the other. Guess what? the array is slower.
Now after all this testing I still use RAID 0 with my two old UDMA33 10GB Fujitsus and I have to admit the loss is not great considering that each drive is now hopelessly slow.
Tried it on two GXP's and all my testing proved that unless you use RAID 0 for what it was designed, you actually lose performance compared to having the two GXPs seperately. i.e Put the OS and program files and you just killed it.
I know I'll take some heat from RAID0 lovers, but go ahead and do some weekend testing and see if I am right or wrong.
IDE RAID 0 is good for A/V editing where sequencial read write matters.
Put the OS on it and you're dead. If you think of it, all the fuss about how multiple drives are beneficiary to performance is cancelled by IDE RAID 0 since you end up with one big disk where all heads are sync. Desperate.
I know Sandra lovers will be excited with the benchmarks but try moving a big file around in the array. I've done tried it:
it takes longer to move/copy a file from one partition of the array to another than to copy/move the same file between the two disks if they are seperate. It makes sence. It's the same as having one disk and two disks. No matter if the one disk is almost 2x faster than the two appart.
Try this. Put the OS on the array and in startup put 10 programs.
do the same on the two disks apart with the OS on the first disk and the programs on the other. Guess what? the array is slower.
Now after all this testing I still use RAID 0 with my two old UDMA33 10GB Fujitsus and I have to admit the loss is not great considering that each drive is now hopelessly slow.
Tried it on two GXP's and all my testing proved that unless you use RAID 0 for what it was designed, you actually lose performance compared to having the two GXPs seperately. i.e Put the OS and program files and you just killed it.
I know I'll take some heat from RAID0 lovers, but go ahead and do some weekend testing and see if I am right or wrong.