Topic: Promise ATA 100/133 Controller for CD?

[rf131]

Hi There,

I have a Dell PowerEdge 500SC that has two built-in ATA100 IDE interfaces.  I have seen some threads about performance issues if you have a CD attached to one of these channels.  I want to do a RAID-1 in software with these channels, and don't want anything else on them to slow things down (initial install with CD on one of the built-in channels is OK though).

I have seen references to the Promise ATA 100/133 PCI controller cards here.  How difficult is it to install and configure the controller for a CD?  If I load up initially with the CD attached to one of the built-in channels, would this make it easier?  Does Anaconda pick this card up in 6.0.1, or do I have to manually install the driver?

Should I follow the Promise instructions after I get done installing the system?

Sorry for the questions, but I did a search and didn't find a whole lot, and some of it was very dated.  I beg your pardon for your time if I overlooked the instructions, I really do try to SFTFI (search for the friendly instructions)!

It really is a bummer I couldn't get the 600SC series... it was the only one I saw that had THREE built-in IDE channels.  I wonder what the install script would do with this?

Thanks,
Kevin

[christian]

My experience with Promise cards has only been on PCs running WinXP but I think the issue will be the same.

I believe they show up as SCSI devices to the OS which may or may not be the issue with WinXP. But what I've seen is a Promise card works fine with HDDs but CD and DVD drives can't seem to work efficiently at all. When I do get them to go they have severe problems with throughput.

It could be driver related with WinXP but somehow I don't think so.

So for my WinXP PC I have my 2 DVD drives and 1 CD drive on the mother board IDEs. I have my Hard drives on the Promise card.

Note you can get other type of true IDE cards which may be easier for you.

Hope this helps a little.

-Christian

[Reinhold]

Kevin,

Not sure I understand you problem... but

use HD1 as master IDE0
use HD2 as master IDE1
use CD/DVD as slave IDE1

This is A.D.2005 don't be afraid that "things will slow down" ... they won´t  
(else why not test YOURSELF with CD cable attached then removed how fast your data transfer is)

If you absolutely want to use a Promise controller then use an "ULTRA" (NOT the pseudo-Raid called "FAST") and just put both HDs on their own channel...)

HTH

Regards
Reinhold

[rf131]

Thanks Reinhold...

I have already done some testing with and without the CD connected.  Either way, the numbers are around:

#hdparm -tT /dev/hda

tbcr:  128 MB in 0.55 seconds = 232.73 MB/sec
tbdr:  64 MB in 1.14 seconds = 56.14 MB/sec


NOTE: /dev/hdc is the second disk and performance is almost identical

I hope these numbers are what would be expected of a 1.1ghz, ATA-100 system.

It appears to make not much of a difference, so I'll just leave it hooked up.

[Reinhold]

Hi rf,

"c"      
BTW: Did you know that you can also test you md-device (the raid device) that way ?!  (just /dev/md0 )
...have a look - useful to kill some other myth

Regards
Reinhold

P.S.: 56MB/s is on the better side of things IDE!
_and_ if you want to debunk even more try them two disks as master/slave on one channel
...not recommended for raid though!

[rf131]

Well,

When connecting the CD, my BIOS didn't find the second disk this time, and I came up in degraded mode.

Time to follow dmays RAID recovery howto...

Got the BIOS to recognize the disk.
Did my raidhotadd on all three partitions, now recovering; this may take awhile (7.3% after 10 minutes).

Nothing like some side issues happening to get you up-to-speed!

Kevin

[rf131]

Well, got the second drive synced and back on-line OK.

Just another issue with the disks I am using and something for those of you building a system...

I have two WD800JB 80GB EIDE drives.  One came original in the PowerEdge 500SC server (3 years old), the other I got on eBay; it was manufactured in May of this year.

There was a significant firmware version difference in the two disks, and that was reflected in the write performance.  I noticed that the newer drive was getting about 15MB better write performance than the old one (read performance was the same).  I didn't take notice of this until after I had hot-added the second one yesterday.  It looks like the firmware version (which cannot be updated   ) makes a difference in the write performance.  The hdparm -tT results on the array itself line up with the slowest disk, which is what I would expect.

The moral of my story here is that even when you get "identical" disks, try to get them at about the same age (newer the better?) and/or have disks that can accept firmware version updates.

Don't get me wrong, I love even the performance that I'm getting, it is just something that those of you out there contemplating the same thing should be on the lookout for.  There may be a potential for a HUGE slow-down with certain drives.

I may just try to get me another newer one and use it with the newer one, and use the older one to swap in and out periodically.  This looks like one of the other methods some of you use to keep yet another backup.

Thanks for all the advice,
Kevin