I'm trying to create a genuinely real-time dashboard to keep track of Amazon Connect agent statuses, but I'm hitting some latency issues. Currently, my setup involves using Agent Event Streams with Kinesis Data Streams, then going through AWS Lambda to send updates to a webhook. Unfortunately, this whole process usually takes between 500 to 2000 milliseconds from the moment an agent changes their status to when it reflects on the UI. Ideally, I'd like to get that down to less than 100 milliseconds if possible. Has anyone successfully managed to get lower latency for agent status updates, and what architecture or solutions did you use? Are there alternatives to Kinesis Data Streams for handling Agent Event Streams, like EventBridge or something else?
2 Answers
What's your actual use case for needing that level of real-time updates?
I rely on the connected signal to trigger recordings when agents join calls, and the dashboard updates when they go on hold. That's why I'm aiming for lower latency.
Give your team at least half a second to adjust. What’s driving the need for such quick updates?
I don’t really see why real-time is crucial here, especially if it’s just for a dashboard. A second of delay seems reasonable since that’s about the time it takes for a person to notice a change on the screen.