If BigQuery is good, there's no reason to. I just assumed a time-series database would be better suited to the workload.