Measuring load against temperature

Measuring server load against temperature

By David Mytton,
CEO & Founder of Server Density.

Published on the 28th November, 2013.

We’ve had our first colocated server running live for just over a week now and have written some custom plugins for Server Density to help us monitor hardware metrics like temperature and power draw. We’ll be releasing these in our plugin directory shortly (there are already 2 plugins for Dell temperature monitoring and Dell fan monitoring) but there are some fun, interesting initial observations regarding load and temperature.

Anyone who has ever run a full screen flash video or Google Hangouts knows that they are good stress tests for your CPU, and after a while the fans start spinning as the CPU usage causes temperatures to increase.

Now we’re graphing that, we can actually see and prove the direct correlation between CPU load and temperature. Look how closely they match each other:

Load against temperature

Load against temp last week

Of course this also means more power is being drawn and we can see a similar correlation between load and power usage:

Load against power

Temperature against power

These graphs are cool to look at, and there’s a purpose behind monitoring metrics of this kind – making sure they stay within acceptable ranges. We want to know if temperatures suddenly go up (could indicate a failed fan or a data center issue), if power suddenly drops to zero on one PSU (again, failure of that PSU) and same for fan speed if it drops too low.

For more information on how you could start monitoring metrics like this, it’s worth trying out our hosted server monitoring software. After all, it’s free for 15 days!

Articles you care about. Delivered.

Help us speak your language. What is your primary tech stack?

Maybe another time