Fixed scaling is easier to implement with longer timeframes
I can see why fixed scales are a problem with the “5 Minutes” view and possibly a problem with the “3 Hours” and “24 Hours” views. But I can’t see why they would be a problem with the “Week”, “Month”, or “Year” views which are so highly aggregated that graph scales vary much less. .
I suspect that most of the design focus is on what happens when fixed scaling the shorter time frames. But independent of the issue for shorter time frames, fixed scaling is worthwhile for many of us users who are as much interested in the long term view.
I have to view the scale and do a mental calculation every time I view the graph.
In terms of the shorter time frames, I want to be able to glance at the graph and see high usage. At present, I have to read the scale off the graph before I have any idea what level of traffic there is.
Normally, I’m only interested in the higher levels of traffic so it is a waste of time having to check the scale to find out whether the graph is showing significant throughput or not.
Here’s an example of how this could be useful. If my son is playing a particular online multiplayer game and I have traffic higher than 5Mbps then he will be impacted. We have a 100Mbps Internet connection and his traffic is low, but the issues arises from local Ethernet collisions and lack of quality QoS on our router. If he calls out, “Are you slowing me down?” then I can glance at a fixed-scale Graph or Mini Viewer and say “Yes” or “No”. At present, I have to look in more detail.
I have to view the graph to know what is really happening in the Mini Viewer
The lack of fixed scaling affects the Mini Viewer even more because there is no scale displayed in that window.
For those reasons, I don’t use the Graph much because it only has autoscaling so every “5 Minutes” or Mini Viewer graph looks much the same at a glance. To make use of that information, I have to first calculate the value on the scaled graph and, second, compare it with my “fixed” value threshold.
In practice, I don’t bother with the short-term graph view because the scaling changes so much. I use the usage view almost exclusively and don’t even bother with short-term time periods. This is much less than ideal because it tells me very little about what is happening right now and I still have to do a manual calculation if I want an hourly figure. But it is much easier to read a number and calculate once rather than having to read a graph and calculate twice.
With autoscaling I always have to check the graph scale even if I had looked at it only seconds earlier. By having a fixed scale, I avoid having to take the additional second or so to read the axis scale every time I view the graph.
What would help everyone is placing the axis scale, or another axis scale figure, at the leading edge of the graph which is where the action is - that is where my eyes are looking. At present, I look at the graph and focus on the far right, I then have to move my focus back to the far top left before returning to the far right. I have to do this every time I view the graph.