Stephen,

Let me say up front that most of my (work) experience comes from SCSI.  But 
the principles hold for ATA drives as well.

One thing that is unique to ATA drives is the fact that you get only 2 disks 
per channel - one master and one slave.  

It is also my understanding that that IDE based disks start a request and 
keep the channel tied up until that request is finished.  On SCSI drives a 
request is issued and the disk can "disconnect" from the SCSI bus until is 
ready fulfill the request.  So SCSI systems have a real advantage there.

RAID:  In general, the two primary modes of RAID used in businesses are RAID 
1 (Mirroring) and RAID 5 (striping with parity).  The reason for this is data 
protection.  RAID 1 has some advantages in terms of write performance over 
RAID 5.  Conversely, RAID 5 can have some read performance over RAID 1.  

Many people pick RAID 5 over RAID 1 because it uses their disk resources more 
efficiently in terms of storage capacity.  Let's take 4 disks.  If you 
protect your data using RAID 1 you will be mirroring and you will get 2 disks 
worth of total storage capcity.  If you use RAID 5 you can get 3 disks of 
total storage capacity.

RAID 1 has some advantages in terms of what has to be done when the disk set 
needs to be rebuilt following a disaster.  If your disk set is RAID 1 then 
you "merely" copy the data from teh surviving disk onto the new mirror.  If 
you are using RAID 5, then the new disk must be buidl with "calculated" data. 
 That is the data from all the other disks (whether it is real data or parity 
data) will be need to be read from the surviving disks.  That total data is 
then used to calculate the data going on to the new drive.

Now for a personal opinion.  I will be building a server for my church.  I 
plan on using ATA100 drives with (probably) the Promise RAID card.  I will 
use RAID 1.

Hope this helps.
-pen

Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com

Reply via email to