Subject: Re: SCSI vs. IDE
To: Paul Hoffman <phoffman@proper.com>
From: Bri <simian@replic.net>
List: netbsd-users
Date: 04/29/2001 14:40:45
 What makes you think that IDE itself has inherent corruption problems?

 Your experiences would lead me to believe that you were working with
faulty hardware. I've never experienced any sort of issues like that.

 SCSI drives are usually geared towards a different market, and the bus
technology is faster in situations where you have lots of simultaneous i/o
requests on a single bus but that dosen't mean that given faulty hardware
it woulden't be corrupting data aswell.

 As far as which IDE controllers work, I've had no problems with the BX
chipsets controllers, or some older VIA's (I've heard the newer ones are
picky). Also the promise and hotpoint work just fine. If you want high
performance, might want to checkout the 3ware I beleve(been mentioned on
this list recentley.)



simian at replic d0t net

On Sun, 29 Apr 2001, Paul Hoffman wrote:

> Greetings again. I bought a Dell refurbished box with the intention
> of using it as a replacement for one of main servers. However, it has
> shown occasional IDE problems just in the few hours of installing
> NetBSD 1.5 on it. So, I'm not going to rely on it and need to buy a
> different box.
>
> In the past, I have noticed some IDE problems when thrashing
> different PC systems, such as untarring pkgsrc.tgz. For instance, on
> this system, copying pkgsrc.tgz to the hard drive and untarring
> caused enough IDE errors for the system to need a full powercycle
> (just resetting the system didn't clear the drive problem and the
> PROM couldn't find the drive). When I untarred the file from the
> CD-ROM, there were no problems.
>
> So, I am lead to believe that IDE isn't the greatest thing in the
> world. My question is almost historic: should I use SCSI instead of
> IDE for reliability? I'm considering getting a Dell 1550 rack system
> that has on-board SCSI instead of the Dell 350, which is IDE-based.
> If not, how do I pick the "right" IDE controllers, particularly
> because every motherboard has its own, and manufacturers switch
> controller chips without warning.
>
> --Paul Hoffman
>