Question

I am facing an issue that I don't believe is novel but none the less am having trouble finding a solution that fits well with our system. We have a constant stream of events going into AWS Kinesis. The way Kinesis works (to my understanding) is it reacts and recalculates some query based on new events. Now what I am trying to do is react to a lack of events. Example: I want to detect when some event (lets say a login event) does not happen for x days in a row.

Now my question is not necessarily about Kinesis (but am definitely open to thoughts on ways one might be able to accomplish this there) but about this concept of "debouncing" events in some stream processor / treating the absence of an event as an event. I am looking for material on how other platforms that handle this, well, handle this at scale. A big use case I can think of is detecting when some log stream fails or stops producing logs which could indicate an issue.

No correct solution

OTHER TIPS

A typical implementation would be a timer that checks that some event has occurred at periodic intervals, and does something if it has not. Sometimes called a watchdog timer.

Licensed under: CC-BY-SA with attribution
scroll top