civileme wrote: >On Sunday 11 November 2001 05:12 am, Tom Brinkman wrote: > >>On Saturday 10 November 2001 10:55 pm, Bill Kenworthy wrote: >> >>>Are there stats to backup the fact that linux software raid is >>>faster than the highpoint chips+linux software raid - I have not >>>seen any articals to that effect >>> >> http://www.mandrakeforum.org/article.php?sid=450&lang=en >> >> I'd want Civileme to respond to your speed query rather than try >>to speak for him. I would add tho, that the win-raid fad that many >>popular boards came out with a while back, requires a pcb design that >>actually introduces many problems. I believe it is at the heart of >>the much (over)hyped VIA-IDE 'bug' problems. Fortunately this >>win-raid kludge fad appears to have died out with mostly clueless >>windoze users being the majority of those who got snookered. Most >>from reading favorable win-raid motherboard reviews on otherwise >>respectable windoze hardware sites. eg, Tom's Hardware, Anand, Ars >>Technica, etc. >> > > >There is nothing to indicate that RAID0 under linux software RAID would be >any slower or faster than RAID0 on one of the HPT or Promise "RAID" >controllers. If you want higher performance on that, you need real hardware. > >Of course these hardware sites reviewed RAID favorably because it gave some >performance hains in Windows over the regular stuff, but realize that these >are basically WinRAID controllers. Obviously if many windows users have the >setup, and feel the need to access their WinRAID partition (yeah >singular--you get _one_ RAID partition with these controllers) and use >dual-boot, a project would be in order to assist them, but notice it is not >by commercial distributors. > What?? That last sentence is absolutely cryptic! What do you mean that you only get "one RAID partition with these controllers"? Once you setup the array, you can partition the drives as though they were only one drive - you can have as many partitions as you can without raid.
>Even the linux-ide project has no personnel >involved. Andre Hedrick wrote quite an interesting letter about this junk on >kernel traffic, more than a year ago, as you can discover from the link Tom >posted. > >For right now, the Promixe RAID controllers are supported for one drive per >channel, as regular controllers, and the HPTs seem to work somewhat better as >regular controllers. > That depends on what controller you're talking about. The fasttrak 66/100 add-on cards work the same as your existing ide controllers; two devices per channel - two channels onboard. When you create the array, you have the choice of 2, 3 or 4 ide devices in that array. > > >There is a supported hardware IDE RAID documented at www.linux-ide.org and it >does work very well indeed. It is a tad on the expensive side, but it could >work wonders in limited space (one device runs two 2.5" notebook size drives >as a RAID0 or RAID1 totally transparent to the system in a single 3.5" >bay--great for a high-reliability thin box... > >Civileme >QA team > > > > > >------------------------------------------------------------------------ > >Want to buy your Pack or Services from MandrakeSoft? >Go to http://www.mandrakestore.com >
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
