I have just set up a squid transperant caching server at the isp that I work at and would like to know how I can tell if I have any bottlenecks.

We set up a machine as a test machine: 667 Celeron, 40 Gig 7200 rpm drive, 384 megs of ram. This machine has been running for 4 days now and is doing much better than we had expected but there is one problem I hadn't thought of. This server was set up to test what bottlenecks we would have and then we are going to build a server to fit the needs we have, How do I tell if the hard drive is a bottleneck?

We have about 1500 Clients going through it right now and are seeing about 20% less bandwidth being used the server is running about 25% proccessor useage with about 700 concurrent connections at a time, what I have been seeing is that it will run steady between 15 and 30% proccessor useage and then about once every 10 minutes it will spike to 98% for about 25 seconds, it is not at any given interval so I don't think it is a cron bothing it (the only thing running is squid) I am wondering if the drive is not able to keep up with the information being read and saved?

Hopefully I haven't given too much info to confuse a very simple answer :-) So my question is how do I find if I have a bottleneck in the system?

Any Suggestions??? How can I tell how much the drive is being used? Not space but throughput kind of like the utility top tells me how much cpu I am using?

You can collect and report a lot of system activity with the command "sar".
This tool is commonly used on all kinds of Unix systems.

For example, if you want some I/O reporting you can use the command "sar -b 1 10", which reports I/O with these parameters 10 seconds long and refreshes every second.

I recommend reading the manpage for this tool, as it is quite extensive.

Sometimes the "vmstat" command is also interesting. This will report how much memory is cached by the system / Squid.

By the way, which Linux distribution do you use?
The RPM file for Red Hat Linux 9 is called "sysstat-4.0.7-3.i386.rpm".

You can check if it's installed with the command "rpm -qi sysstat".

If you use a Red Hat distribution, you may want to check out which services are active in which runlevel, you can use the command "chkconfig --list |more".

(
You can find the homepage of the author of the Linux implementation of 'sar' here : "http://perso.wanadoo.fr/sebastien.godard/"
)

Good luck!