RHQ/JON Chart Types Decision -- Input needed

Elias Ross genman at noderunner.net
Fri Nov 2 07:46:17 UTC 2012


On Tue, Oct 30, 2012 at 8:59 PM, mike thompson <mithomps at redhat.com> wrote:

>
> There has been many opinions on this issue and we wanted further feedback
> from the community on their thoughts. So please, reply back with your
> thoughts...

I think this might require some customization to do right, there's
nothing I saw in the PDF I really cared for or seemed appropriate.

My big requirement is having the chart representation match the data
that was captured by the database. 1:1 is ideal.

For near-time data, say the past 24 hours or so, I would expect to see
points joined by a line, where each line actually represented a value
taken by the agent. If there are really too many values for that range
(say, 200+) then aggregate, fine. Please don't simplify the data if
possible: There may be details that are important to capture (spikes,
etc.)

For aggregate data that has 'min max avg' values, I would expect to
see that data represented by three lines, where the middle line
represented the average value, and the others were dotted (or lighter)
with the same color representing the range for that data. I would
avoid making an 'area' graph out of the data, since the most important
is the average and what's usually of interest to users is the average
alone. (Maybe emphasize the area if the mouse hovers over a min/max
lines.)

Having area charts makes sense when the values are closely related,
say CPU percentages or number of data packets received and sent.

You might do a very simple candlestick where the average was the line,
and a light vertical line for that point showed the range in time.
This might look a little strange.

I think a fair bit of user customization might be required depending
on the data. The visualization you choose should be remembered by RHQ.


More information about the rhq-devel mailing list