Just wondering if anyone else has been down this route?

I currently have Seagate 80GB:IDE:7200RPM Barracuda drives (x2).
Im planning an upgrade of the server for more than one reason but in relation to the Hard Drives....
Has anyone upgraded from 7200 IDE drives to SCSI/SAS 15K drives at all?

On paper they should be twice as fast?
Does anyone have any experience of what its like in reality or have a site they could show me that is hosted on these drives?
I realise that many things come into play as limiting factors etc, network, connectivity, rooting etc etc but im not sure how else its possible to see if there would be any improvement
Thanks in advance.

Depending on your server's IDE drive controller it most likely will not be able to support upgrading to 15k SCSI drives (and definitely not SAS drives), although there are a few exceptions. I'd be happy to take a look at your server to determine whether or not the drives could be upgraded to SCSI, please just PM me your customer ID.

My experience going from ide to sata to scsi and sas drives is that you won't see the difference until push comes to shove (kind of like upgrading computers now... when you upgrade to a new faster one it's often not immediately clear that it's so much faster, until you do something that really needs the power. Then going back to the slow one it's immediately obvious how slow it was.) With servers, I found the iowait frequently spiking and causing very noticable ~1 second lags if I got a traffic spike while doing log rotations or cpanel backups. Going to scsi/sas I found the io system much more "robust" in terms of handling multiple tasks in stride so backup/rotation operations didn't affect the foreground tasks as much. We're seeing sata drives get more capeable than they used to be, but when looking at a new server I'll always choose at least 10k sas/scsi if it's within my budget.

What does your existing iowait look like on your current server (or wa in top) during your heavy load time of day? (if you're running linux)

Or how does iostat look (optional handy command, installed with yum install sysstat on rhel5)

@ Brooke
Thanks, ive already checked, the current server cant support these drives

@ Jeff
Thanks very much for your imput here, iowait/iostat is no problem. I was more interested in peoples thoughts as to the drive speed differences, is upgrading from 7200 to 15,000 noticeable?

If the iowait is minimal, then so will the resultant change. iowait is basically how much does it have to wait on account of the data access being slow.

if its never particularly high (be it as part of a spike, or average levels) then its not worth it. if the spikes are problematic, or the base level is high, then defiantly go for it

if you are having performance problems and the iowait isnt high, then you will need to better benchmark and work out where the bottleneck is.

remember no matter how fast a system is there will always be a minimal time for something to execute.

Just to reinforce what Martyn says, figure a 7200 RPM drive might have 4 ms of latency and a 15k drive might have 2 ms of latency. If you're 50 to 100 ms away from the server (or even if you're connected via console) you're not going to notice 2 ms. But if you multiply that by millions of requests per day, it makes a difference to the overall performance and it means the slower drive is going to start becoming the bottleneck sooner.

Thanks for the imput here guys!
Having done a bit more research it seems there really isnt that much of a benefit.

Yeah I thought upgrading would help me but support said there would not be much difference.

When you reach the point of needing more disk io/second it will be a nice option to have on the table though for the future.