10-26-2010 07:14 AM - edited 10-26-2010 07:52 AM
I will be building a replacement Remote Desktop Server soon. We currently serve 15-25 concurrent users on an old Windows 2003 server with 4GB RAM, 3-disk RAID5 config (2 logical drives: C & E). Performance often suffers and I am not sure if that is a function of the processor, RAM, or disk config, but I suspect it is mostly a combination of processor and RAM, and a little bit disk config. There are no plans to add users and I highly doubt we will ever have more than 25 users on at one time.
Now for the replacement server.
It will be an RD120 that we already have. Windows 2008 R2 64bit Remote Desktop Edition. RAM is currently at 6GB and expandable. The trouble I am having is trying to determine my RAID config. From what I have read, a RAID5 with 2 logical drives is not a high performance config, but that is all I know and have ever used.
The C drive on this machine will hold all the user profiles (default). The E drive on this machine will contain the various programs:
Antivirus, MS Office, proprietary apps, etc. Typically files that the users will be working on will be located on other servers.
For my disk config, I would like high performance and good redundancy. But if a disk does fail, I must have hot swap capability, and during that rebuild, I don't care if performance suffers. I am OK with 1 disk fault tolerance per array, in other words, no particular need for hot-spare at this time. I have gathered that this is likely the best scenario:
For C: RAID1 (mirroring / 2 disks)
For E: RAID1 (mirroring / 2 disks)
Is this a suitable high-performance disk configuration for a terminal server? Thank you so much for your insight.
Solved! Go to Solution.
10-26-2010 10:58 AM
in your situation, RAID 1 would be good since it doubles read speeds and carries a low overhead on the controller due to its simplicity. access speed drops to single-disk performance during a rebuild.
if you don't need a lot of storage capacity and have the budget, you could consider an SSD for both C: and E:. these could be imaged periodically with Windows Server Backup onto a SATA drive to provide redundancy. this would reduce access speeds and increase reliability. the backup drive could even go to a RAID 1 array or network drive. in the event of an SSD failure, it would be a matter of re-imaging the SSD using a backup.
Share your input on the Retro ThinkPad Time Machine