@mackuba.eu On each one the catchup speed slightly varies over time, but they make similar patterns, but appearing at a different wall-clock time. So it means the ups & downs depend not on the server load or websocket speed, but on what data was in the firehose at a given seq (larger events, larger records?…)