is it though? not for time series workloads with billions of rows it isn’t