I want to show a "success"/"failed" value on a chart in a compact space. Basically I have various processes that run every day. I want to show in a compact space the status history of each process.

The purpose of the chart would be to alert the user to problems and provide a launch for them to drill down into a particular process that they see had failed recently (even if it was recently successful).

Here's an example of an idea I just made in paint:

enter image description here

  • 1
    The example you made in paint is largely identical to Amazon's AWS Service Health Dashboard, though Amazon has a separate copy of the current day's data at the top. This allows faster access to detailed current status information, as that's usually more important than historical status information. – Brian Sep 17 '13 at 19:31

I think you could use sparklines as a compact visualization tool.
enter image description here
This view is not precise, but it allows to view errors distribution and total error count for each process. More precise information is displayed in process specific screen.

Light grey bar is an observation time window (a week or two, etc.). Dark line is an error occured within observed period. It's placed according to its occurrence time: today's error is placed rightmost an so on. So you could perform simple visual analysis of recently failed processes.

Probably you'll pay more attention to the faulted processes, which have more dark lines closer to the right side (more recent): enter image description here – this that I've called the distribution.

Take a look for the animation, each frame is a day:
enter image description here

  • Interesting. I don't understand how that is distributed. What decides where in the light grey bar the line appears? – richard Sep 17 '13 at 22:40
  • @RichardDesLonde, I've updated the answer. – Alexey Kolchenko Sep 18 '13 at 6:36
  • Ah! That's a really great visualization! I think I will use it. What I like about it is that each grey area could be for a different time frame as well. – richard Sep 18 '13 at 7:06

I'm wondering if the success/fail values in your graph have the same properties as those generated by a continuous integration server:

  • Successful builds are essentially boring
  • Sequential failed builds usually have the same cause

If so, you could bundle "runs" of the same status together in the chart, something like this: BinaryTimeline

(The righthand five bubbles show the same sequence as your original demo.)

These could be stacked in a number of ways to cover your 3+ processes; the exact technique would depend on whether the processes were coupled or independent.

  • Whoa! Mind blown. Thanks. Variations on that could be awesome. – richard Sep 18 '13 at 4:28

Alexey's idea is clean and straightforward- made me think of this image from my Fitbit flex. Sort of what Alexey was proposing, but with some additional information in the form of a hover- giving specifics. enter image description here

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.