Monitiq Heat Maps first appeared in the Oz release. They were enhanced for Pike with further changes planned for future Monitiq releases. This is the first in a series of articles aimed at identifying when and how to employ these useful charts, and presents a couple of easy to understand examples to introduce their use.
Monitiq heat maps allow you to investigate large amounts of data in a number of useful ways. The most obvious of these is looking for time-based patterns in usage.
Of course, these patterns may be deduced from examination of the standard, linear graphs which Monitiq has always provided. So, let’s examine a couple of distinctly different time-based patterns using traditional graphing and also using a Monitiq heat map, to see why using heat maps is quicker and more intuitive.
Using the Group Aggregation View, we’ll look at a number of servers. In particular, focusing on network use over five weeks. The initial GAV gives us a graph like the following, which summarises data from each server over the 5 week period:
Other than a general deduction that network activity is higher at some times than others, there’s not a lot to say. Of course, we can zoom in to any single day to get the full detail:
From this we can see significant activity around 6am, but patterns (other than a background “ripple”), if they exist are impossible to see. We could look at each day of the 5 week period and we may be able to find out more. Fortunately, the Monitiq heatmaps make this easier.
There are a number of inbuilt heat maps to look at, but we’ll begin with the default one which is initiated when clicking on the appropriate button. At the moment, these buttons are only available within the GAV, and are to be found in the “sigma” row(s):
Clicking on the button brings up the default “Day vs Hour” heat map:
Note that Monitiq heat maps are written to make use of WebGL for interactive, 3D use. If your browser, OS or hardware are not WebGL capable, you will see a simple table view of the heat map:
Assuming you have WebGL, you will be able to use mouse gestures (click and drag) to move the heat map around in 3D space:
The immediate observation we can make is that there is significant network activity every day between around 5am and 6am. Investigation into the cause of this is beyond the scope of this article, suffice it to say that this is due to large database backup over the network.
Both the 3D and flat, table views provide hover information on cells:
Our next example looks at a group of application servers delivering intranet pages within an organisation.
A clear weekly cycle of network I/O (together with some missing data). Let’s look at the heat map:
and from the “other direction”:
So, most people seem to start work around 8am, and finish around 4pm, but there is some activity at all times.
So, there we are. A quick introduction to Monitiq Heat Maps. In later articles I’ll look at some of the other fun stuff we can do.