1

We've got a Dell PowerEdge T510 which has 8 disk slots in the front. It is currently utilising 6 of these slots, and the disks are controlled by a PERC H700 controller. The disks are currently running in the following configuration

  • 2x500GB in RAID 1 (OS partition)
  • 4x2000GB in RAID 5 (file storage partition)

Considering that this server is our domain controller which hosts the Active Directory, I'd hate to see the server go down due to disk failures. Thus, I'd like to upgrade the server with disk fault tolerance in mind (and not storage capacity or speed). However, I'm not quite sure what my options are.

I was thinking of simply adding 2 more 500GB disks and convert the OS partition from RAID 1 to RAID 6, as this partition would be the most critical. We do have backups of important data on the RAID 5 partition, and it can be recreated within about 24 hours, but the OS partition would need more as it would require me to do a complete restore of Windows Server including Active Directory from backups.

However, I still acknowledge that RAID 5 has the habit of "acting up" during rebuilds, and I would of course love to convert this one to RAID 6 as well. However, the chassis only has 8 slots.

What are my options to secure this system with regards to disk fault tolerance?

Thanks!

UPDATE:

  • Even though not stated initially, we do of course keep backups of the OS partition (Active Directory). I know that RAID is not a backup system.
  • A second DC is not an option due to budget and licensing restrictions on SBS 2011.

The question was aimed towards my disk fault tolerance options. In other words, I'm looking for a good RAID setup.

sbrattla
  • 1,456
  • 3
  • 26
  • 48

2 Answers2

2

To answer your question directly, the thing that can best improve your system's tolerance to hard drive failure is to use your extra hard drives to hot spare parts. With such a low disk count, the chances of having two simultaneous drive failure is low enough that you can rely on backup. Having a hot spare, however, will allow the system to recover much more quickly (and safely) from a single drive failure. You'll need two spares, though: one of each size.

I have to add, however, that it sounds like your heading in the wrong direction here. First, you need to understand that RAID is not a backup system. There are many things that can go wrong with your RAID that will cause total data lose: manipulation error, controller failure, viruses, fire/water on the server itself. You need to use a real backup system.

Second, the best way to increase redundancy for an AD is not to improve the disk subsystem or have better backup but to add a second domain controller, if possible at a different physical location. It will improve uptime much more than anything you will do for a single server setup.

Stephane
  • 6,382
  • 3
  • 25
  • 47
  • Thanks for your input on hot spares. It is a good idea. Only thing is that the rebuild process will be just as "vulnerable" to errors on one of the other disks. Not sure why you got the impression that I use RAID as a backup system. I do keep separate backups, but recovering systems from backups takes time. I'd rather secure the system up to the point where I never would have to use the backups - even though of course I keep them. – sbrattla May 07 '13 at 10:10
  • @sbrattla "We do have backups of important data on the RAID 5 partition" that's why – Stephane May 07 '13 at 10:15
  • 1
    Really, the proper way to secure AD (beside backups which are an different scenario) is to add additional domain controllers. – Stephane May 07 '13 at 10:18
  • Yeah, it might be understood that way...but I also wrote that "...would require me to do a complete restore of Windows Server including Active Directory from backups." which relates to the OS partition. Anyway, no need to argue over that one. Adding a second DC isn't really an option "money wise". We're also running SBS 2011 Essentials which probably due to the licensing doesn't allow this. – sbrattla May 07 '13 at 10:21
1

Considering that this server is our domain controller which hosts the Active Directory, I'd hate to see the server go down due to disk failures. Thus, I'd like to upgrade the server with disk fault tolerance in mind (and not storage capacity or speed). However, I'm not quite sure what my options are.

The "best" options to protect Active Directory are to stop fiddling around with disks in the one server and to do the following:

  1. Create at least one more DC in that domain.
  2. Backup both DCs with a backup system that can at least backup & restore the system state of the DCs.

Regardless of how many DCs you have and how many disks you have in them and how you've arranged those disks, if you haven't backed up AD then, with respect, you've done precisely nothing worth doing to protect it.

These are the simplest ways to improve the availability of your domain and the safety of the data within your AD.

Rob Moir
  • 31,664
  • 6
  • 58
  • 86
  • Thanks. However, we're a company with 10 people. Just buying an extra server might not be an issue in a large company, but it just is not doable from a "money wise" perspective in this case. I have to make the best of the resources I have, and that is why I need to choose the best option - even though it might not be the ultimate option. – sbrattla May 07 '13 at 10:15