PDA

View Full Version : Help Me: XServe best practices for set up?




yellow
Nov 7, 2007, 12:05 PM
So, here is what we have:

We have an establishd AD domain, which users authenticate against.
We have a department that has data stored on a Windows 2003 server running Services for Macintosh, and they just about all have illegal characters in the files names, and therefore cannot be moved to a straight SMB share because of data/filename dependencies. We inherited this department from an outside group.
Our Windows server admins won't be allowing Services for Macintosh on the new servers, so we purchased an XServe for them so I can do AFP shares for their existing data.
The XServe has a RAID card in it and 3x 750GB HDs.
I have no practical experience with OS X Server beyond the Server Essentials class, which doesn't really cover the conceptual stuff, only the practical applicaitons of OS X Server basics.

Here is what I am trying to accomplish:

Users connect to this server to get to the AFP share. But they can use the same username and password from the AD domain.. automagically. I'd prefer not to have to recreate each account manually. Is this possible to do? Do I bind the OD to AD, thereby allowing the users that exist in AD to use that authentication to get to their AFP shares on this XServe? Perhaps I only need to kerberize this XServe to accomplishg this end? Can I even use AD to authenticate users on the OD for AFP shares only?

Use the RAID card and HDs to their most effective use. Should I have 1 RAID (1) Set that includes 2 disks and store their data on that RAID Volume and have the OS on a separate volume? Or should I have 1 RAID (5) Set that includes all 3 disks? Currently I am leaning towards the latter, this department needs to be able to get to their data.


Thanks for any help.

EDIT (issue 1): The AD/OD Integration paper by Joel Rennich seems to indicate that I can accomplish this, but I'd very much like to set this up correctly from the start and not have to go back and redo everything because I was simply fiddling.

EDIT (issue 2): When I create a RAID (5) Set with 3x750GB drives, my expectation is that this would only give me ~698GB of data storage, since it's mirrored across the 3 drives, but it's telling me that I have 1.34GBs of available space when I create a RAID volume. That doesn't seem right. Should I just tell it to use 698GB of space?



jeremy.king
Nov 7, 2007, 12:42 PM
EDIT (issue 2): When I create a RAID (5) Set with 3x750GB drives, my expectation is that this would only give me ~698GB of data storage, since it's mirrored across the 3 drives, but it's telling me that I have 1.34GBs of available space when I create a RAID volume. That doesn't seem right. Should I just tell it to use 698GB of space?

Nothing to worry about. With RAID 5, the capacity is equal to the size of the smallest drive x (number of drives - 1)

yellow
Nov 7, 2007, 02:42 PM
Thanks for the info. This is a tough choice to make. :)

trainguy77
Nov 8, 2007, 12:50 AM
I would go with RAID 5 in this situation for a two main reasons.

1. Either way you get the same amount of space. But with RAID 5 you can partition it so you don't waste so many gigs on the system. Who uses 750gb for the system anyway?

2. A drive failure won't take your system offline. With your first option a drive failure (could be the system drive) would take your server offline until you rebuild your system, which could take some time. With RAID 5 ANY of the 3 drives fail you just replace that one drive and your good to go.

yellow
Dec 7, 2007, 11:04 AM
For those that might come across this at a later date:

It is definitely possible for a user to connect to shares on the XServe via AFP/SMB and authenticate via Kerberos with the XServe simply being part of the "Connected to a Directory System". The ACLs for shares can be pointed at users/groups within the AD structure.

Oh, and in the end, I opted for RAID5 (1.34TB volume) and I'm very pleased with the results.