I have some Dell PowerEdge 2850's with SCSI drives. They each have two. I have always configured them to be RAID 1 or RAID 5 when there is more than 3. I want to know a little more about the SCSI portion of it. By enabling the SCSI adapter to utilize SCSI intead do regular RAID, what are the benefits besides speed? If a drive fails, can you just simply pull it out plug another in and your good? Is SCSI redundent like regular raids? I have I just want to make sure I understand this. If I need more space, can I just add a drive? All of these will run CentOS.

Greetingz!

When you say this, do you mean in the firmware of the controllers (hardware-based RAID), or in the OS itself (software-based RAID)?

A speed benefit depends entirely on what RAID type you're using (hardware vs software, which is usually negligible with modern hardware), and what level (RAID 0, RAID 1, RAID 5, RAID 0+1, RAID 10).


Depends on the controller and the server. Hot-pluggable is hot-pluggable. Normally, if the controller supports hot-pluggable, and the drives are in little removable caddys, you're good.
The inverse being if you have to open the case and take a screwdriver to a drive, you should probably power-off the server first.

SCSI is just a protocol type that defines how devices interact on a bus (like IDE or SATA, or FC-AL). SCSI itself is *not* redundant. For that you need to depend on a RAID level (other than RAID 0).
I think you'd find the wikipeida page on RAID pretty handy.

Typically, it's not that easy. You'll need to look at a logical volume manager of some sort (like Veritas StorageFoundation if you want a commercial product with support).


Ah! In this case, you'll want to read-up on the mdadm command, and take a look at the documentation for LVM2 (logical volume manager).

Example: Here's a layout that is very common amongst servers.

Thanks for the great response. Let me answer these for you.

In the firmware of the BIOS and the controllers.

Yes they are the Caddys. Dell PowerEdge 2850 and 2600

I had looked at this. There is a lot of good information here. It seemed that it set an array of disks so that you can use several smaller drives and then seem to look like one disk. It seemed to me to be like a RAID 1 or 5 or something like that. That's what made me think it may be redundant as well. Also you could maybe add space...?

[/QUOTE]

I do need to understand linux partitions better. Windows servers no problem.


So it seems I am better off in my case to just go with RAID 1 since I would have "some" redundancy. When I installed CentOS on one machine with two SCSI drive setup in the BIOS it made all 1 large drive. I made another exact model system and BIOS settings and it sees the drives as two. And I understand why. I guess one of the things I do not grasp is, Why have a server that can have 8 SCSI drives hot-swappable all small drives running linux or Windows. and any of those fail, your done without a tape or something. It seems if you had 8 IBM drives all the same model and had something flawed seems there is more to go wrong and potentially all at the same time. I just don't want to loose the data on these machines and I want to configure them for the long run and I do not plan to use tapes. In your opinion should I just simply setup a RAID in the BIOS and set it up for RAID 1 for this setup?

Let me add to this. The system BIOS allows me to configure it for RAID or SCSI ok. Does this mean operating as a SCSI protocol, can you not use a RAID configuration? I