Click here to see the SAS code.
Click here to see the example.

I wrote the original version of this chart over 20 years ago, when I worked
in the unix support group here at sas.  We used this kind of chart to plot
cpu data (and load averages, etc) for up to 100 machines on one page, and 
it was very useful in finding overloaded machines, and machines with run-away
jobs hogging the cpu (for example, when you see several red lines side-by-side).

This one uses a trick similar to barcpu -- once again
I'm using gplot & skipmiss to plot a bunch of colored line segments.
In this case, they are all of the same length, and form a 
"spectrum" chart of cpu usage data.  

Along the left axis are the nodenames (hostnames) of several computers,
and all of the colored lines to the right of that name represent the 
cpu usage of that host at that time (see timestamps along the bottom).
Each line segment represents an hourly average cpu usage.  

As with the barcpu, this technique allows you to visualize a *lot* of data 
on a single chart - but you have to give up the charttip/drilldown, since 
that is not supported on gplot line segments.

Back to Samples Index