A message queue is an architectural element of the Stampede monitoring. This diagram shows how different types of messages could be inserted and pulled off the queue to modularize the functionality. See how, for example, the calculation of durations for various workflow events could be split off as its own element and the results could then be used by various anomaly detection (or visualization, or other analysis) components. Of course, other anomaly detection components, such as those detecting errors, could operate on the event stream directly -- or even a combination of the two.

  • No labels