Did you know that multiple data streams can share one Y-axis in an instrument if the metric is configured correctly?
Let's say you are using CollectD and want to build a graph of the different memory metrics. If you just add the metrics to a graph without configuring the metrics' attributes it can look like this:
The axes are stealing valuable graph real estate and it is hard to detect anomalies due to the different scales.
To change that, all you need to do is configure the attributes for each metric:
The first parameter (Y-Axis Label) not only names the Y-axis, it is also used to determine if data streams can share the Y-axis when you are building an instrument. The second parameter (Tool-Tip Label) determines what is displayed in the tool tip. Notice how I used the display transform to convert the bytes sent by collectd to megabytes.
After setting the units for all the collectd.memory.memory.* metrics making sure that all metrics share the same unit, the graph on the dashboard looks like this: