Let's now examine the performance characteristics of our new data. In the first query, against the home_status_update_ids table, we are performing a range query across a single partition; we know that this is an extremely efficient query and in the form we should strive to always use in our application. The second query, however, accesses several partitions, one for each status update author in the result set. If we're using a page size of ten, we might read updates from as many as ten partitions, although it might be fewer if more than one of the updates comes from the same author.
This is an improvement over our previous approach: the number of partitions that we need to access is no longer proportional ...
