Load is loosely tied to the CPU. Load is more so an average based on the number of actively running processes in the process list of the OS.
The three load numbers are based on the averages for a given time frame. The first number is the past minute, the second number is the past 5 minutes and the last number is the last 15 minutes (I think the last one is that, but it could be 10 as well).
So if you have one job that has been in the run queue for the past minute, you will have a load of 1.0 for the first load number, 0.20 in the second number, etc.
Things that keep a process in the run queue can be CPU, IO, or something system related (waiting on a child process perhaps).
Load is a good indication of what the server is doing overall, but not really a good indication of what the CPU or HDs are doing because it can be skewed to a certain extent.
To answer you original question: "How can I see what the HD is actually doing?" run the following command from a shell as root:
vmstat 2
Pay close attention to the columns under "-----io----". The numbers under the column show how much disk IO is occurring, according to the kernel. BI is "Blocks In" or reads, BO is "Blocks Out" or writes.
That will give you an understanding of what is being read and written to your HD.
As for "over working the drive"... That generally isn't possible. HDs are one of the only systems in a computer that are designed to fail from the moment they are installed. Some fail long before they should, and some fail long after they should have. Google has a a white paper out that describes their observations as to HD failure in their datacenter operations. I highly recommend anybody that operates a server to read it, as it is very enlightening.
Here is a link to the article:
http://labs.google.com/papers/disk_failures.pdf
Facebook
Twitter
