To avoid high volume metrics causing high CPU/memory overhead and dropped metrics, we want to start accumulating individual metric emissions.
So instead of sending out a buffer that looks like this:
[logback.events+1, logback.events+1, logback.events+1, .... logback.events+1, important.metric+1]
We accumulate these to:
[logback.events+99, important.metrics+1]
To avoid high volume metrics causing high CPU/memory overhead and dropped metrics, we want to start accumulating individual metric emissions.
So instead of sending out a buffer that looks like this:
We accumulate these to: