Hi I'm getting frequent messages that my Linode is having a lot of disk IO. How can I analyse this with a view to reducing it if necessary?

Typical message I get via email:
Your Linode, my-VPS, has exceeded the notification threshold (1000) for disk io rate by averaging 1442.34 for the last 2 hours.

You want to figure out what your server is doing when these messages are generated. If it's just, say, that your weekly backup is running, you probably don't have anything to worry about.

How can I view that?

Try iotop during a period of high IO.

I am running CentOS 5.5 and I am unable to install iotop on it. Please help

The CPU usage and transfer quota are measured in percent, and so they make some kind of cognitive sense.

But the 1000 ops/sec doesn't have a meaning without context. Is that a lot? Or, busy but not overwhelmed?

My cronjob consumes ~30K OPS - it's not like you're almost at your limit, it'll manage.

To be nice for the other people on your physical node it'll be good to watch with iotop though, to see what's causing it.

Unfortunately, you can't transfer IOPS into a useful percentage since there are so many factors that can determine the max IOPS. Read versus write, sequential versus random, 4KB versus 32KB, etc.