Choosing a Time Series Database for High Cardinality Aggregations