table point is that the scan queries each achieve higher throughput
than their time series query counterparts for the same time range.
This point is not surprising, since the scan queries are faster. Finally, for a given number of clients, the throughput for each query
type decreases as the time range of the query (and hence the amount
of data scanned) increases.
Figure 13 shows that the response time for the queries increases
in proportion to the number of clients, as expected.