Hi Civileme Merry Christmas
Thanks for your competent advice and time spent. I would only consider true RAID, hardware RAID in practical use. In parallel, I would not skip software RAID to envisage my knowledge in this respect. Maybe in some days late it will replace hardware RAID. B.R. Stephen At 12:43 PM 12/23/2001 -0900, you wrote: >On Sat, 2001-12-22 at 23:04, Stephen Liu wrote: > > Hi All People, > > > > Merry X'mas > > > > I have some further RAID questions to ask : > > > > 1) Can ATA-133 hard drive be used on RAID controller available on market > > such RAID 0, 1, 0+1 etc. with cable for ATA100/ATA133. No special > designed > > controller is needed ? > > > >Yes, but there is no support for the fake hardware RAID in 8.1. Read >support is sort-of there to allow compatibility with reading a Windows >RAID but there is no support for using the software RAID in the PROM on >the RAID "controller" that is actually a BIOS extenstion. > >You folks don't know you are paying $30-60 extra to a hardware vendor >for a BIOS extension PROM? Or that the software in there is protected >by kaw from reverse engineering so that being compatible with the >WinRAID controller is at the mercy of whatever scraps of infor the >manufacturer will give us? > >Do yourselves a favor. Don't buy the fake RAID unless you really need >it for windows. Even then, real hardware RAID is preferable and >available here: > >http://www.arcoide.com/ > >And for info on the open source RAID project to provide some >compatibility with the WinRAIDs, go here: > >http://people.redhat.com/arjanv/pdcraid/ > > If you want to know which controllers support full 133 cable speed, >check this: > >http://www.linux-ide.org/chipsets.html > >Of course you can use a 133 drive on a lesser controller--no harm done. >You just cannot get full speed out of it. > >Finally, if you are considering buying new IDE equipment, be aware of >which of it is actively supporting linux by checking here: > >http://www.linux-ide.org/endorsements.html > >In particular, one of the brands not listed is a source of continuing >problems just to get the disk geometry right and should never, ever be >used above ATA / 33 on any machine though they rate themselves for 66, >100 and 133 without the hardware on board the drive to meet the >standard. And if you ask them they will tell you in no uncertain terms >that they don't support linux. > >Linux has its own software RAID which covers modes 0, 1, 4, and 5, 3 is >best left to real hardware. > >For most linices, the drivers are the same, right out of the kernel, >plain vanilla and written to the _standards_ defined by the hardware >manufacturers. More than one hardware manufacturer meets those >standards with a combination of corner-curring dodges on the hareware >and their own third-party drivers (for windows) to make up for the >deficiencies of the hardware. Naturally they work only with windows >because no one who is economizing will see any cost-effectiveness in >writing for a system that has only 5% market share on desktops,(and the >folks who use servers for real business buy something else anyway). One >even offers a detection service direct to windows users to pre-warn you >when your drive is about to fail. You can search under "Data Lifeguard" >to find it, but it works only with Windows. We have a similar feature >by activating the SMART technology on your drives, but just local to >your computer. > >Now, insofar as ATA/xx is concerned, sometimes the highest possible xx >is not a good idea. Before 8.2, I will issue another versiuon of >drakopt to run the tests to find the best speed for your drives. There >are many cases where a higher speed _reduces_ performance because the >noise causes more repeat requests, and there are many situations where >ATA/66 actually outperforms ATA/100 on an individual machine, and some >cases where ATA/33 outperforms ATA/66 (most severe I saw was a WD200AB >on a MaxData computer set up by RH to run ATA/66 which provided >1.86Mb/s. At ATA/33 with optimizing settings determined by DrakOpt, it >was getting 9.62 Mb/s.) Remember that higher speeds mean smaller >signals in terms of total charge, and the energy in these signals are >now small compared with a secondary cosmic ray. > >Please, I have answered this question enough. Next time someone else >answer with a link to the archives pointing to this or any of the other >articles. > >Civileme > > > > 2) RAID 0 > > If I have 6 discs and the controller has only 2 channels. How to make > > connection ? 1 master and 2 slaves for each channel ? Or need a > > controller with at least 3 channels > > > > 3) RAID 1 > > a) Question same as RAID 0, about discs connection : > > in order to achieve two concurrent separate reads per mirrored pair or two > > duplicate writes per mirrored pair ? > > b) What does it mean "One Write or two Reads possible per mirrored pair" > > > > 4) RAID 3 to my understanding : being very high read and write data > > transfer rate > > a) Controller design is fairly complex : is it awailable on market ? Not > > ordered made component ? > > b) Connect question for 3 hard discs > > c) What does it mean "Low ratio of ECC (Parity) disks to data disks means > > high efficiency" ? > > > > Thanks in advance. > > > > B.R. > > Stephen Liu
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
