Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am looking at this new Lenovo ThinkServer RD240 I got. Since we are using NAS for file storage, and most of our servers are more function/application oriented (not requiring lots of data) I am trying to decide how I should order my drives.

Basically, I can put 8 250GB 7.2k drives in a Raid 5 array and still have plenty of storage for my virtual machines. My questions is this...

Should I go Raid 5 with lots of small hard drives, or Raid 5 with 3 or 4 larger drives. Or maybe even Raid 1 with 2 really large drives?

I believe RAID 5 with small discs will give the best performance, just not sure what drive configuration to spend my dollars on. Any advise?

FYI. The virtual machines on this server will be a domain controller, terminal server, small web and sql server for low performance database, possibly a windows 7 admin virtual machine for me to remote access the network via RDP from.

share|improve this question
up vote 1 down vote accepted

The size of drive doesn't affect performance- 8 250GB drives will go the same speed in terms of IO/s and MB/s as 8 2000GB drives, so long as the spindle speeds (7.2k RPM) are the same.

Also, regarding raid level, if you have space to spare (and it sounds like you do), consider raid 10. You lose half of your usable capacity to overhead, but you can compensate for that by buying larger drives. With raid 10, you can lose as many as half your disks before losing data, the performance for reads (which should be 70% or more of your IO) will be as good or better than a raid 5 with the same number of drives, because you have the same number of disks being used for each read both ways. Also, sparing in a replacement for a failed drive on raid 5 takes longer since the server needs to rebuild the spare from "parity", which means doing a lot of math and a lot of reads on your existing drives.

You don't want to use a small number of large drives because doing local virtualization can cause a disk performance bottleneck, and the only way to increase disk performance is to increase the number of "heads" (or "spindles") in your raid.

In short:

  • Always try to have the largest number of drives possible
  • Pick the size of drives based on how much storage space you'll need
share|improve this answer
Thanks for the great feedback! I wonder if you can install xenserver on the same partition as the vm storage without affecting performance? – Jessica Moyer Oct 1 '11 at 5:25
Even if you don't install it on the same partition, all partitions are striped across the same raid array. If that raid is capable of doing 800 IO/s, then the sum of your storage IO/s coming in on all your partitions can get up to 800 IO/s. – Basil Oct 2 '11 at 3:02
The 800 I/O thing is new to me. What does it mean and how can I measure it? – Jessica Moyer Oct 2 '11 at 23:48
It was an example, but each drive in a raid contributes some number of IO/s (for small block IO) and MB/s (for large block or sequential IO). The better the raid controller, the closer to the aggregate performance of the disks you'll get. For example: if you have 8 drives capable of doing 100 IO/s each, then your performance for non-cached IO will be at most 800 IO/s. It could be less if you have a less efficient raid controller, though. – Basil Oct 2 '11 at 23:55

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.