About 3 weeks ago when I was building a new server, I put 2 identical SATA drives in the server (500GB each), each capable of 60MB/s transfer when clocked.  I had a fair amount of data (250GB) to share on these 2 drives.  I normally would opt for a raid mirror (hardware based mirror), however I have this data backed up else where, so I decided that I would combine the disks using a span to make a 1TB partition.

I transferred the data to the partition in 2 days and really thought nothing of it because I was busy with other stuff.  However when I got back to college with this server, I found that most of my local file operations would take 10 times longer than they should take.

Therefore, I decided to ditch the span.  To copy all of the data off the span took 4 days to an external hard drive (capable of 20MB/s).  To copy from the external to a simple dynamic volume took a little under 2 hours.  This proves to me that spans just plain suck.  I’m now using just 1 disk and use the other as a weekly backup disk.

I did discover after a little research that Windows treats spanned and stripped volumes differently although they both combine physical disks to make bigger partitions.  Windows NTFS (NT File System) has a 2TB physical limit that it can handle, and NTFS starts becomming inefficient after 500GB to 1TB.  Since a span treats 2 physical disks as 1 physical disk, this explains the slow down.

However, I did learn that a stripped volume will treat each disk as a separate disk and combine the storage into 1 logical partition.  I have yet to try this with another disk and I may comment on this if I ever try using a stripped volume.

The bottom line is, you will have a major performance problem under server 2003 x64 bit if you opt for a spanned dynamic disk, not to mention 20 to 30 server crashes while getting your data off to undo it.

Tags: 2003, disk, dynamic, dynamic volume, server, span, x64
Posted in Hosting / Server Administration

Leave a Reply

Click here to cancel reply.

Name (required)

Mail (will not be published) (required)

Website

CAPTCHA Code *