Warning: count(): Parameter must be an array or an object that implements Countable in /home/iteanz/public_html/wp-includes/post-template.php on line 317
Maximize the Streams Graph card. You can also enlarge it by using the resize handle at the bottom right of the card. Enlarge it just enough to show the entire graph. Move it to another position and remove other cards as you see fit.
Review the graph. The graph is familiar from the Instance Graph in Studio, though it represents information in slightly different ways. It labels every stream with the tuple rate, and indicates operator health by a colored dot. As in the Instance Graph, relative tuple rate sets the thickness of the arrow. Usually the Throttled stream, at 40 tuples per second (give or take a few), is the thickest, but every so often the Observations stream, normally at zero, exceeds it. You observed the same behavior in the Flow Rate Chart.
The Streams Graph is an alternative to the Summary card to detect trouble (unhealthy PEs) and to identify bottlenecks that can affect throughput performance. A bottleneck is an operator that limits the flow of tuples, usually because it cannot process any more tuples per second with the CPU cycles it has. If you do the optional section on back-pressure at the end of Lab 4, you will see that the Throttle operator is the cause of congestion that builds up over time.