Right now we use an aggregated CPU utilization metric to compute system load (on Redshift). This aggregated value is not enough to scale Redshift because of the different compute node roles in the cluster. We should leverage the NodeID dimension to get more detailed metrics.
Right now we use an aggregated CPU utilization metric to compute system load (on Redshift). This aggregated value is not enough to scale Redshift because of the different compute node roles in the cluster. We should leverage the
NodeID
dimension to get more detailed metrics.