Thanks for speedy reply!
Have done a little bit of homework on this (stress the _little_
)
Highpoint's, Promise's, and 3X's upper-end controllers all seem to run into the USD$100's. Dual-processing systems can easily add much more USD$$'s and OS/CPU processing req's to this cost
RAID is a Redundant Array of Inexpensive Disks, not a Redundant Array of Expensive
CPU's+controller, yes!?
Two identical lower-end Highpoint controller's as mentioned above TOGETHER(!) cost less than USD$100.
Quote:
I do not suggest using software RAID for level 5 because it takes a lot processing power to handle read and write transactions
|
Reviewed a bit of the PCGuide's RAID 5 info
(
http://www.pcguide.com/ref/hdd/perf/...gle_Level5.htm)
and it pretty much agrees with what you write.
IMHO, main reason to do a RAID 5 system would be to have good fault-tolerance with three ATA133 drives at low-cost, even with much lower performance from SW-RAID system.
Question is in having one controller with this "hardware" RAID 0+1 level, or in having two such controller's set up into a low-level (but real) software RAID 5; which is really better in terms of fault tolerance at lowcost (even w/ terrible performance) ??
Seems from reading PCGuide's RAID info, that having two cheap add-on PCI controllers available with on-card BIOS chips would actually allow for slightly better performance w/ an OS-managed SW setup in my own scenario than one controller, no?
Two PCI ATA133 controller's would allow for four separately processed channels for making a RAID 5 array with one or two extra "master-jumpered" striped disks (making a total of 3-4 disks on array)
Even with one disk on each of the two controllers, duplexing fault-tolerance would be better (a little faster?) than mirroring fault-tolerance, the latter with only one PCI controller card..
-nycace36