Justifying the need to compare
To compare data, we can of course display a first chart and then a second chart and compare visually the 2 generated charts in 2 separate windows. This works if differences are obvious. Sometimes differences are more awkward especially when axis don't have the same scale.
An illustration of this: below an example of an Oracle RAC database with several nodes and we want to compare the activity of node1 and node2 for a given day:
Without a compare function we could see this for node1:
and this for node 2:
All y axis have a different scale and it's very difficult to compare something.
Instead of this, KAIROS gives the capability to have this view:
or even this view:
KAIROS allows to draw several plots within a unique chart. They are sharing either the Y axis or the X (time) axis.
In the charts above, the yellow area shows the CPU consumption on each node. Sharing Y axis has the effect to have a better feeling of what the difference is. Sharing X axis is useful to perceive a global impact on a particular period in time.
How to compare
In the example above, we are proceeding like this:
a) create a node who will be the cluster view
b) create within CLUSTER_VIEW a new node who will be the comparison between NODE1-2015-03-25,3 and NODE2-2015-03-25,3
c) drag NODE1-2015-03-25,3 over the created node while pushing the "Ctrl" Key
Once the drag & drop is complete, the yellow chip turns green
d) drag NODE2-2015-03-25,3 over the created node while pushing the "Ctrl" Key
After this step, if we open the "green" node, we should see this:
The node has a type "C" (like compare) and it has 2 producers.
The node is now ready to be browsed. All charts available against producers will be available in the node dedicated to the comparison.
How to share Y axis or X axis ?
Answer: through the settings options.
There is an option named "Plot orientation". When the selected value is "horizontal", plots are sharing the Y axis, and when the selected value is "vertical", plots are sharing the X (time) axis.
Is it possible to compare aggregated data ?
Indeed, you can create aggregated data or consolidated data and at the end compare this data with other nodes. But comparison is the last step: you cannot aggregate or consolidate data from "compare" nodes.
Mnemonic way to associate keyboard keys with "aggregate" or "compare"
"Aggregate" begins with letter "A" like the "Alt" key
"Compare" begins with letter "C" like the "Ctrl" key