Am again in a design dilemma on how to architect solution regarding platform events limits.
From the API documentation, even though I can publish 2.4Million Events in a day, the Comet D client(Java Middleware) will only get 50K events pushed. So after my cometD reaches 50K events, it has to poll manually, again providing last received event ID to get/replay newly published event. So manually polling defeats the purpose of using the event-driven architecture. I agree that events will be present in eventBus for 24 hours, but then what's the point of manually polling after every few minutes after reaching 50k. How can I make my client know that events wont be pushed now and I have to start polling?
Can anyone shed some light about this? And how to optimize the polling? Src : https://developer.salesforce.com/docs/atlas.en-us.platform_events.meta/platform_events/platform_event_limits.htm